diff --git a/agent-context-kit/agents-md-validator/index.html b/agent-context-kit/agents-md-validator/index.html new file mode 100644 index 0000000..f7d48ea --- /dev/null +++ b/agent-context-kit/agents-md-validator/index.html @@ -0,0 +1,67 @@ + + + + + + + + + AGENTS.md Validator — ACKit 0.5.4 + + + + + + + + + + + + + + + +

AGENTS.md Validator

One repository, many agents: AGENTS.md (Codex), CLAUDE.md (Claude Code), GEMINI.md (Gemini) and Copilot instructions plus nested scopes. When files disagree, agents follow stale or contradictory guidance and waste runs.

ACKit resolves every instruction file into a provider-aware graph with scope, precedence and provenance, then reports conflicts, duplicates, shadowing and dead files — before your agent reads them.

ackit instructions --explain
+ackit optimize --explain
codex/AGENTS.md → claude/CLAUDE.md (provenance: frontmatter applyTo)

Try it without installing: npx --yes @cynrath/agent-context-kit@latest readiness

Instruction GraphOptimizeGitHubnpm

+ + + + diff --git a/agent-context-kit/ai-agent-readiness/index.html b/agent-context-kit/ai-agent-readiness/index.html new file mode 100644 index 0000000..f138bd4 --- /dev/null +++ b/agent-context-kit/ai-agent-readiness/index.html @@ -0,0 +1,71 @@ + + + + + + + + + AI Agent Readiness — ACKit 0.5.4 + + + + + + + + + + + + + + + +

AI Agent Readiness

Is your repository actually ready for Codex, Claude Code, Cursor, Copilot or Gemini — or will the agent burn context on missing instructions, leaked secrets and bloated includes? Guessing costs tokens; measuring is one command.

ACKit scores instructions, security, context efficiency, task hygiene, skills and policy on a deterministic 0–100 scale with weighted renormalization, strict gates and baselines.

ackit readiness
+ackit readiness --strict
+ackit readiness --fail-below 80
Readiness  89/100  ██████████████████░░  (threshold 0 — pass)
+  Instructions          90/100
+  Security              90/100
+  Context Efficiency    70/100

Try it without installing: npx --yes @cynrath/agent-context-kit@latest readiness

ReadinessCLI ReferenceGitHubnpm

+ + + + diff --git a/agent-context-kit/benchmarks/index.html b/agent-context-kit/benchmarks/index.html index ac48030..3cc534c 100644 --- a/agent-context-kit/benchmarks/index.html +++ b/agent-context-kit/benchmarks/index.html @@ -6,7 +6,7 @@ - Benchmarks — ACKit 0.5.2 + Benchmarks — ACKit 0.5.4 @@ -16,7 +16,7 @@ - + @@ -52,12 +52,16 @@ VS Code Security Benchmarks - Migration + Migration + AGENTS.md Validator + AI Agent Readiness + Context Optimization + MCP Repository Tools GitHub

Benchmarks

ACKit benchmarks cold scan, warm scan, incremental scan, memory, throughput, context-pack construction, instruction-graph construction and cache behavior.

node benchmarks/run.mjs --classes small --out /tmp/out
 node benchmarks/check-thresholds.mjs
- + diff --git a/agent-context-kit/checkpoints/index.html b/agent-context-kit/checkpoints/index.html index b947be8..a4f36d6 100644 --- a/agent-context-kit/checkpoints/index.html +++ b/agent-context-kit/checkpoints/index.html @@ -6,7 +6,7 @@ - Checkpoints — ACKit 0.5.2 + Checkpoints — ACKit 0.5.4 @@ -16,7 +16,7 @@ - + @@ -52,7 +52,11 @@ VS Code Security Benchmarks - Migration + Migration + AGENTS.md Validator + AI Agent Readiness + Context Optimization + MCP Repository Tools GitHub

Checkpoints / Resume / Handoff

Checkpoints snapshot task state with deterministic resume context and handoff packs, so work survives provider switches and interruptions.

ackit checkpoint create --next-objective "wire gate" --next-path src/core --next-command "pnpm test" --next-expected "green"
@@ -61,7 +65,7 @@
 ackit checkpoint export --format json --out .ackit/reviews/handoff.json
 ackit checkpoint import .ackit/reviews/handoff.json
 ackit task resume

Portable handoff v2

ackit.handoff.v2 binds task and workflow state, the checkpoint record, evidence presence, the verification binding (state and bundle digests) with the latest verdict trust summary, checkpoint staleness at export time, the canonical status snapshot (verbatim blockers plus derived next actions), a redaction manifest, and provider-neutral resume instructions. Imports validate against current disk state and stay read-only: fresh handoffs render resume context, moved state refuses with VERDICT-STATE-STALE, v1 markdown refuses with HANDOFF-V1-UNBOUND, malformed files with HANDOFF-INVALID, and unknown tasks with HANDOFF-TASK-UNKNOWN.

- + diff --git a/agent-context-kit/cli/index.html b/agent-context-kit/cli/index.html index 906ff97..b2a4620 100644 --- a/agent-context-kit/cli/index.html +++ b/agent-context-kit/cli/index.html @@ -6,7 +6,7 @@ - CLI Reference — ACKit 0.5.2 + CLI Reference — ACKit 0.5.4 @@ -16,7 +16,7 @@ - + @@ -52,7 +52,11 @@ VS Code Security Benchmarks - Migration + Migration + AGENTS.md Validator + AI Agent Readiness + Context Optimization + MCP Repository Tools GitHub

CLI Reference

The CLI favors stable terminal and JSON output so the same checks can be used interactively and in CI.

ackit init --dry-run
@@ -79,7 +83,7 @@
 ackit diagnostics bundle --out ./ackit-diag.zip --redact-check
 ackit dashboard --host 127.0.0.1 --port 0 --open
 ackit mcp serve

Exit codes

0 success, 1 threshold failure, 2 usage/configuration, 3 environment, 4 security, 5 internal error.

- + diff --git a/agent-context-kit/context-optimization/index.html b/agent-context-kit/context-optimization/index.html new file mode 100644 index 0000000..80a7b7b --- /dev/null +++ b/agent-context-kit/context-optimization/index.html @@ -0,0 +1,68 @@ + + + + + + + + + Context Optimization — ACKit 0.5.4 + + + + + + + + + + + + + + + +

Context Optimization

Every duplicated guidance block and oversized context file is tokens your agent pays for on every run. ACKit finds the waste with evidence — file paths, token-waste estimates and provenance — and proposes safe, reviewable fixes.

Pair the advisor with budgeted, provider-aware context packs so each agent gets exactly the context it needs.

ackit optimize --explain
+ackit pack --profile codex --max-tokens 50000
# Optimize Report (1 suggestion(s))
+- **[low] budget-overrun** (low): 867 file(s) did not fit the 20000-token budget

Try it without installing: npx --yes @cynrath/agent-context-kit@latest readiness

OptimizeProvider ProfilesGitHubnpm

+ + + + diff --git a/agent-context-kit/dashboard/index.html b/agent-context-kit/dashboard/index.html index 9da55a6..e437466 100644 --- a/agent-context-kit/dashboard/index.html +++ b/agent-context-kit/dashboard/index.html @@ -6,7 +6,7 @@ - Dashboard — ACKit 0.5.2 + Dashboard — ACKit 0.5.4 @@ -16,7 +16,7 @@ - + @@ -52,12 +52,16 @@ VS Code Security Benchmarks - Migration + Migration + AGENTS.md Validator + AI Agent Readiness + Context Optimization + MCP Repository Tools GitHub

Dashboard

The dashboard binds to localhost by default and provides local views for scan results, graph, readiness, tasks and policy.

ackit dashboard --port 0 --open
 ackit report serve ./report.html --port 0
- + diff --git a/agent-context-kit/diagnostics/index.html b/agent-context-kit/diagnostics/index.html index 2511ece..f75b7bb 100644 --- a/agent-context-kit/diagnostics/index.html +++ b/agent-context-kit/diagnostics/index.html @@ -6,7 +6,7 @@ - Diagnostics — ACKit 0.5.2 + Diagnostics — ACKit 0.5.4 @@ -16,7 +16,7 @@ - + @@ -52,12 +52,16 @@ VS Code Security Benchmarks - Migration + Migration + AGENTS.md Validator + AI Agent Readiness + Context Optimization + MCP Repository Tools GitHub

Diagnostics

Diagnostics describe the local environment, configuration, cache, policy and task state without exposing plaintext secrets.

ackit diagnostics --json
 ackit diagnostics bundle --out ./ackit-diag.zip --redact-check
- + diff --git a/agent-context-kit/drift/index.html b/agent-context-kit/drift/index.html index a0cea3c..8354f57 100644 --- a/agent-context-kit/drift/index.html +++ b/agent-context-kit/drift/index.html @@ -6,7 +6,7 @@ - Drift — ACKit 0.5.2 + Drift — ACKit 0.5.4 @@ -16,7 +16,7 @@ - + @@ -52,13 +52,17 @@ VS Code Security Benchmarks - Migration + Migration + AGENTS.md Validator + AI Agent Readiness + Context Optimization + MCP Repository Tools GitHub

Drift

Drift detection reports deterministic findings when implementation diverges from the declared plan, with a managed pre-commit gate for the active workflow task.

ackit drift check
 ackit drift check --ci
 ackit drift check-active --ci
- + diff --git a/agent-context-kit/evidence/index.html b/agent-context-kit/evidence/index.html index d40ebff..cadd51b 100644 --- a/agent-context-kit/evidence/index.html +++ b/agent-context-kit/evidence/index.html @@ -6,7 +6,7 @@ - Evidence — ACKit 0.5.2 + Evidence — ACKit 0.5.4 @@ -16,7 +16,7 @@ - + @@ -52,14 +52,18 @@ VS Code Security Benchmarks - Migration + Migration + AGENTS.md Validator + AI Agent Readiness + Context Optimization + MCP Repository Tools GitHub

Evidence

The evidence registry links acceptance criteria to typed proof. Criteria sync from the task document; checkbox state is never copied, so implementation is never confused with verification.

ackit evidence sync
 ackit evidence show
 ackit evidence verify --criterion AC-001 --type test --ref "vitest run"
 ackit evidence validate
- + diff --git a/agent-context-kit/getting-started/index.html b/agent-context-kit/getting-started/index.html index ad86f4a..fb49014 100644 --- a/agent-context-kit/getting-started/index.html +++ b/agent-context-kit/getting-started/index.html @@ -6,7 +6,7 @@ - Getting Started — ACKit 0.5.2 + Getting Started — ACKit 0.5.4 @@ -16,7 +16,7 @@ - + @@ -52,16 +52,20 @@ VS Code Security Benchmarks - Migration + Migration + AGENTS.md Validator + AI Agent Readiness + Context Optimization + MCP Repository Tools GitHub -

Getting Started

ACKit requires Node.js 22 or newer. Installation is global or via npx; repository analysis runs locally.

Install

npm install --global @cynrath/agent-context-kit@0.5.2
+  

Getting Started

ACKit requires Node.js 22 or newer. Installation is global or via npx; repository analysis runs locally.

Install

npm install --global @cynrath/agent-context-kit@0.5.4
 ackit --version
 ackit --help

First pass

ackit init --dry-run
 ackit readiness
 ackit scan --ci
 ackit instructions --explain
-
+
diff --git a/agent-context-kit/github-action/index.html b/agent-context-kit/github-action/index.html index 9477c3e..bca2585 100644 --- a/agent-context-kit/github-action/index.html +++ b/agent-context-kit/github-action/index.html @@ -6,7 +6,7 @@ - GitHub Action — ACKit 0.5.2 + GitHub Action — ACKit 0.5.4 @@ -16,7 +16,7 @@ - + @@ -52,7 +52,11 @@ VS Code Security Benchmarks - Migration + Migration + AGENTS.md Validator + AI Agent Readiness + Context Optimization + MCP Repository Tools GitHub

GitHub Action

Use the official action for repeatable repository checks in CI with least-privilege permissions.

permissions:
@@ -62,13 +66,13 @@
     runs-on: ubuntu-latest
     steps:
       - uses: actions/checkout@f548e57e544e1ff5a4c46bf1e1b8685f8e4a348a
-      - uses: Cynrath/agent-context-kit@v0.5.2
+      - uses: Cynrath/agent-context-kit@v0.5.4
         with:
           command: scan
           args: "--json"
           fail-threshold: high
           upload-sarif: "false"
-
+
diff --git a/agent-context-kit/index.html b/agent-context-kit/index.html index 0d107b3..73e45d5 100644 --- a/agent-context-kit/index.html +++ b/agent-context-kit/index.html @@ -6,17 +6,17 @@ - AgentContextKit — ACKit 0.5.2 - + AgentContextKit — ACKit 0.5.4 + - + - + @@ -52,10 +52,14 @@ VS Code Security Benchmarks - Migration + Migration + AGENTS.md Validator + AI Agent Readiness + Context Optimization + MCP Repository Tools GitHub -

AgentContextKit 0.5.2

Turn repository context into something agents and humans can inspect. ACKit combines readiness scoring, instruction graphs, context packs, policy-as-code, security scanning, tasks, MCP, dashboard and editor tooling without sending repository data to a hosted analysis service.

Getting StartedCLI ReferenceGitHubnpm

Install

npm install --global @cynrath/agent-context-kit@0.5.2
+  

AgentContextKit 0.5.4

Turn repository context into something agents and humans can inspect. ACKit combines readiness scoring, instruction graphs, context packs, policy-as-code, security scanning, tasks, MCP, dashboard and editor tooling without sending repository data to a hosted analysis service.

Agent context and repository readiness for Codex, Claude Code, Cursor, Copilot and Gemini — validate agent instructions, scan repositories, optimize context and enforce evidence-backed workflows. Offline-first, deterministic, no telemetry.

Getting StartedCLI ReferenceGitHubnpm

Try it without installing: npx --yes @cynrath/agent-context-kit@latest readiness — one command scores any repository 0–100 with actionable findings.

Install

npm install --global @cynrath/agent-context-kit@0.5.4
 ackit --version
 ackit --help

Quickstart

ackit init --dry-run
 ackit readiness
@@ -64,7 +68,7 @@
 ackit sync --check
 ackit pack --profile codex --max-tokens 50000
 ackit dashboard --port 0 --open

Core capabilities

  • Agent Readiness — explainable 0–100 scoring with strict gates and baselines.
  • Instruction Graph v2 — provider-aware scope, precedence, provenance, conflicts and duplicates.
  • Context Packs — deterministic ranking, token budgets and per-file manifests.
  • Policy Packs — machine-checkable repository rules.
  • Scanning + SARIF — secrets, connection strings, entropy, absolute paths, CI pinning and redacted evidence.
  • Tasks — docs-first single-active workflow and completion gates.
  • Workflows + Intent — staged lifecycles with gates, committed intent docs with fingerprints.
  • Evidence + State-Bound Verification — criteria-to-proof registry, state-bound bundles (ackit.verification-bundle.v2), persisted bound verdicts (ackit.verdict.v2) with verifier independence and replay/staleness refusal.
  • Checkpoints + Portable Handoff v2 — snapshots, verification-bound portable handoff (ackit.handoff.v2), read-only import with stale refusal.
  • Canonical Status — read-only task/stage, blockers, verification and checkpoint freshness, next actions (ackit.status.v1).
  • Provider / Surface Parity — CLI, SDK, read-only MCP, Action and VS Code project the same status snapshot.
  • Containment Hardening — link-aware output guard with symlink/junction escape repair; MCP stays read-only.
  • Trust-Flow Demo — reproducible offline proof from missing evidence to second-process resume.
  • Drift + Policy v2 — deterministic drift findings, risk-tiered autonomy and review policy.
  • Roles + Skills — portable role contracts, managed open-standard skills.
  • Managed Sync — content-driven reconciliation of ACKit-owned assets after upgrades.
  • MCP + SDK — local integrations for agents and tooling.
  • Dashboard + VS Code — local inspection surfaces for readiness, findings, graph, tasks and policy.
-
+
diff --git a/agent-context-kit/instruction-graph/index.html b/agent-context-kit/instruction-graph/index.html index 3f1025b..d41059f 100644 --- a/agent-context-kit/instruction-graph/index.html +++ b/agent-context-kit/instruction-graph/index.html @@ -6,7 +6,7 @@ - Instruction Graph — ACKit 0.5.2 + Instruction Graph — ACKit 0.5.4 @@ -16,7 +16,7 @@ - + @@ -52,12 +52,16 @@ VS Code Security Benchmarks - Migration + Migration + AGENTS.md Validator + AI Agent Readiness + Context Optimization + MCP Repository Tools GitHub

Instruction Graph v2

Instruction Graph resolves repository guidance from Codex, Claude, Gemini, Copilot and shared files. It tracks scope, precedence, provenance, shadowing, duplicates and conflicts.

ackit instructions --explain
 ackit instructions --provider codex --for src/app.ts --json
-
+
diff --git a/agent-context-kit/intent/index.html b/agent-context-kit/intent/index.html index 460754c..4caa3ab 100644 --- a/agent-context-kit/intent/index.html +++ b/agent-context-kit/intent/index.html @@ -6,7 +6,7 @@ - Intent — ACKit 0.5.2 + Intent — ACKit 0.5.4 @@ -16,7 +16,7 @@ - + @@ -52,14 +52,18 @@ VS Code Security Benchmarks - Migration + Migration + AGENTS.md Validator + AI Agent Readiness + Context Optimization + MCP Repository Tools GitHub

Intent

Intent documents capture the goal of a work item before implementation, with machine-path-independent fingerprints and secret-gated validation.

ackit intent new "ship offline packs"
 ackit intent list
 ackit intent validate
 ackit intent fingerprint
-
+
diff --git a/agent-context-kit/llms-full.txt b/agent-context-kit/llms-full.txt index e81b0a2..c43475f 100644 --- a/agent-context-kit/llms-full.txt +++ b/agent-context-kit/llms-full.txt @@ -1,4 +1,4 @@ -# AgentContextKit 0.5.2 +# AgentContextKit 0.5.4 # ACKit — AgentContextKit @@ -20,7 +20,22 @@

Turn any repository into an agent-ready repository.

-

Evidence-backed completion · Independent verification · Cross-agent handoff · Deterministic drift · Provider-independent contract · Offline-first
Instruction graph · Skills · Scanning · Context packs · Tasks · Policy · Readiness · MCP
Offline-first · Deterministic · Task-first — ackit on Node ≥ 22

+

Agent context and repository readiness for Codex · Claude Code · Cursor · GitHub Copilot · Gemini · MCP
Validates agent instructions, scans repositories, optimizes context and enforces evidence-backed workflows.
Offline-first · Deterministic · No telemetry — ackit on Node ≥ 22

+ +

Try it in 30 seconds — no install:

+ +```powershell +npx --yes @cynrath/agent-context-kit@latest readiness +``` + +

One command scores your repository 0–100 with actionable findings.

+ +- Detect conflicting agent instructions (`AGENTS.md` / `CLAUDE.md` / provider scopes) +- Find unsafe repository context (secrets, connection strings, absolute paths) +- Measure repository readiness across instructions, security and context efficiency +- Reduce wasted agent context (duplicate guidance, oversized files, low-value includes) +- Enforce evidence-backed task completion with independent verification +- Generate provider-aware context packs (Codex / Claude / Copilot / Gemini / generic)

Quickstart → · @@ -159,7 +174,7 @@ OpenCode, Copilot, Gemini, Cursor, Cline, any MCP-capable agent). 🩺Diagnosticsackit diagnostics bundleackit.diagnostics.v1 + deterministic bundle-manifest.json (sha256 + redaction count), 5-secret [REDACTED] proof ⚡Benchmarksbenchmarks/run.mjs7 deterministic fixtures, 8 metrics (coldScanMs/warmScanMs/incrementalMs/peakRssMb/filesPerSec/packMs/graphMs/cacheHitRatio), median-of-3 🔌SDK v1import { scanRepository } from "@cynrath/agent-context-kit"sideEffects:false, type:module, exports {".","./mcp"}, AbortSignal <200ms, AckitError -🧩VS Codeextensions/vscode0.5.2, Cynrath, lints Linters, onStartupFinished, readiness tree + Problems ACKITxxx + status-integrated Tasks view (canonical blockers/next actions, no mutation), <2MB VSIX — Marketplace: Cynrath.ackit-vscode +🧩VS Codeextensions/vscode0.5.4, Cynrath, lints Linters, onStartupFinished, readiness tree + Problems ACKITxxx + status-integrated Tasks view (canonical blockers/next actions, no mutation), <2MB VSIX — Marketplace: Cynrath.ackit-vscode 🤖MCPackit mcp serveOfficial SDK stdio, 16 read-only tools (incl. canonical ackit_status, workflow status, intent, checkpoint, verification bundle, drift, roles), 5 resources, 4 prompts, InMemoryTransport cancellation 🔐Trust-Flow Demodocs/guides/demo-trust-flow.mdreproducible offline proof: missing proof blocks → bound independent verdict enables → mutation stales → fresh re-verification restores → handoff exports → second process resumes (tests/e2e/trust-flow-demo.test.ts) @@ -174,11 +189,11 @@ OpenCode, Copilot, Gemini, Cursor, Cline, any MCP-capable agent). ```bash # global (recommended) npm install --global @cynrath/agent-context-kit -ackit --version # 0.5.1 +ackit --version # 0.5.4 # one-shot, pinned -npx --yes @cynrath/agent-context-kit@0.5.1 --version -npx --yes @cynrath/agent-context-kit@0.5.1 --help +npx --yes @cynrath/agent-context-kit@0.5.4 --version +npx --yes @cynrath/agent-context-kit@0.5.4 --help # from source pnpm install --frozen-lockfile && pnpm build @@ -308,7 +323,7 @@ More: [`docs/architecture/overview.md`](docs/architecture/overview.md) ### 🤖 GitHub Action -Official `Cynrath/agent-context-kit@v0.5.1` (SHA-pinned for high-assurance): +Official `Cynrath/agent-context-kit@v0.5.4` (SHA-pinned for high-assurance): ```yaml permissions: @@ -318,7 +333,7 @@ jobs: runs-on: ubuntu-latest steps: - uses: actions/checkout@f548e57e544e1ff5a4c46bf1e1b8685f8e4a348a - - uses: Cynrath/agent-context-kit@v0.5.1 + - uses: Cynrath/agent-context-kit@v0.5.4 with: command: scan args: "--json" @@ -328,7 +343,7 @@ jobs: # optional: upload findings via actions/upload-artifact@v4 ``` -Inputs `command/args/fail-threshold/upload-sarif`, outputs `findings-json/sarif-path`, annotations/SARIF 2.1.0/job summary, least-privilege. Guide: [`docs/guides/ci.md`](docs/guides/ci.md) · Action: [`action.yml`](action.yml) +Inputs `command/args/fail-threshold/upload-sarif`, outputs `findings-json/sarif-path`, annotations/SARIF 2.1.0/job summary, least-privilege. Shareable local output: `ackit scan --format markdown` (copyable summary) or `--format sarif` for code scanning. Guide: [`docs/guides/ci.md`](docs/guides/ci.md) · Action: [`action.yml`](action.yml) --- @@ -367,13 +382,14 @@ ESM-only, `sideEffects:false`, `AbortSignal` cancellable, no `process.exit`. Ref ### 🧩 VS Code -Extension is **published on the VS Code Marketplace** — [`Cynrath.ackit-vscode`](https://marketplace.visualstudio.com/items?itemName=Cynrath.ackit-vscode) (`0.5.1`, `<2MB` VSIX, offline-first, no telemetry). +ACKit is available on the VS Code Marketplace as +[`Cynrath.ackit-vscode`](https://marketplace.visualstudio.com/items?itemName=Cynrath.ackit-vscode). From source: ```bash pnpm --filter vscode build -vsce package # → ackit-vscode-0.5.2.vsix +vsce package # → ackit-vscode-0.5.4.vsix ``` Features: readiness tree, Problems `ACKITxxx`, graph “instructions for current file”, tasks/policy/optimize, palette `Refresh/Show Graph/Optimize/Diagnostics`, file watcher debounced, no telemetry. @@ -424,7 +440,7 @@ See [`CONTRIBUTING.md`](CONTRIBUTING.md) for the docs-first workflow. ### 🔖 Versioning -Development: **`0.5.2`** on `master` · Latest stable: **`0.5.1`** · [Changelog](CHANGELOG.md) · [Releases](https://github.com/Cynrath/agent-context-kit/releases/latest) · `latest → 0.5.1` via OIDC Trusted Publishing with provenance. Stable pointer: [`release-state.json`](release-state.json). Legacy `.NET/NuGet 1.0.0-rc.1` at `258918b` is frozen. +Development: **`0.5.4`** on `master` · Latest stable: **`0.5.4`** · [Changelog](CHANGELOG.md) · [Releases](https://github.com/Cynrath/agent-context-kit/releases/latest) · `latest → 0.5.4` via OIDC Trusted Publishing with provenance. Stable pointer: [`release-state.json`](release-state.json). Legacy `.NET/NuGet 1.0.0-rc.1` at `258918b` is frozen. --- @@ -447,6 +463,61 @@ All notable changes to ACKit (`@cynrath/agent-context-kit`) are documented in th This project follows Semantic Versioning. +## [0.5.4] - 2026-09-22 + +Release-recovery patch (no runtime feature behavior beyond 0.5.3; no new +commands, reporters, or services). Closes the partial 0.5.3 release: +npm 0.5.3 shipped while Marketplace stayed 0.5.2 because the +tag-trigger workflow assumed an unpublished `vsce publish --oidc` +capability (`@vscode/vsce` 4.0.0 has no `--oidc`; verified via +`publish --help`). + +### Fixed + +- Release workflow no longer attempts automated `vsce publish --oidc`. + A fail-closed manual Marketplace gate verifies + `Cynrath.ackit-vscode` live == release version BEFORE npm publish, + so npm can never run ahead of Marketplace again (TASK-0094, + ADR-0033 amendment; ADR-0023 manual path restored as normal). +- Contract tests lock the manual-gate reality (`--oidc` absent, + `--skipDuplicate` rejected, ordering: marketplace-live gate before + npm publish, no automated `vsce publish`). +- Version coupling preserved: root == extension == tag == release + (0.5.4). Stable pins + `publishedStable` stay 0.5.2 until all live + surfaces verify, then flip to 0.5.4 in the same release plan. +- `v0.5.3` preserved untouched as the partial-release marker; 0.5.3 + history below unchanged. + +## [0.5.3] - 2026-09-22 + +Organic discovery/activation patch (no runtime feature behavior beyond +0.5.2; no new commands, reporters, or services). + +### Changed + +- npm metadata: description now names the multi-provider scope (Codex, + Claude Code, Cursor, Copilot, Gemini, MCP-based coding agents) with + instruction validation, repository scanning, context optimization and + evidence-backed workflows; keywords add `agents-md`, + `agent-instructions`, `claude-code`, `codex`, `cursor`, + `github-copilot`, `gemini`, `context-engineering`, + `repository-readiness`, `developer-tools` (existing keywords kept). +- README first viewport: single install-free try CTA + (`npx --yes @cynrath/agent-context-kit@latest readiness`), provider + line, outcome-focused bullets and offline/deterministic trust note; + deep technical content kept below with anchors intact. +- README/Action docs surface the existing shareable path + (`ackit scan --format markdown`, `--format sarif`, Action SARIF/job + summary); `action.yml` description improved for Marketplace discovery + (`name: AgentContextKit` unchanged, inputs/outputs compatible). +- VS Code README opens with a 30-second onboarding path (open folder → + automatic readiness → findings) plus the install-free terminal + equivalent; extension CHANGELOG gains the `0.5.3` section. +- Getting-started guide gains the install-free readiness check. +- Install/Action pins keep pointing at published stable `0.5.2` until + publication; the stable pointer flips to `0.5.3` after the tag-trigger + release verifies (ADR-0029). + ## [0.5.2] - 2026-09-06 VS Code Marketplace content parity patch (no runtime feature behavior diff --git a/agent-context-kit/llms.txt b/agent-context-kit/llms.txt index 4d8d963..a2fa45b 100644 --- a/agent-context-kit/llms.txt +++ b/agent-context-kit/llms.txt @@ -1,10 +1,10 @@ # AgentContextKit -> AgentContextKit (ACKit) 0.5.2 is an offline-first deterministic toolkit for turning repositories into agent-ready working environments. +> AgentContextKit (ACKit) 0.5.4 is an offline-first deterministic toolkit for turning repositories into agent-ready working environments. ## Documentation -- [AgentContextKit](https://cynrath.github.io/agent-context-kit/): Offline-first deterministic toolkit for agent-ready repositories. +- [AgentContextKit](https://cynrath.github.io/agent-context-kit/): Agent readiness and repository context toolkit for Codex, Claude Code, Cursor, Copilot, Gemini and MCP-based coding agents. Offline-first, deterministic. - [Getting Started](https://cynrath.github.io/agent-context-kit/getting-started/): Install and initialize AgentContextKit. - [CLI Reference](https://cynrath.github.io/agent-context-kit/cli/): ACKit commands, options and exit behavior. - [Readiness](https://cynrath.github.io/agent-context-kit/readiness/): Explainable ACKit repository readiness scoring. @@ -34,6 +34,10 @@ - [Security](https://cynrath.github.io/agent-context-kit/security/): ACKit offline-first threat model and redaction guarantees. - [Benchmarks](https://cynrath.github.io/agent-context-kit/benchmarks/): Deterministic ACKit performance benchmarks. - [Migration](https://cynrath.github.io/agent-context-kit/migration/): Upgrade AgentContextKit to the current release. +- [AGENTS.md Validator](https://cynrath.github.io/agent-context-kit/agents-md-validator/): Validate AGENTS.md and CLAUDE.md instructions; find conflicts, duplicates and scope problems. +- [AI Agent Readiness](https://cynrath.github.io/agent-context-kit/ai-agent-readiness/): Score any repository 0–100 for AI coding-agent readiness with explainable deductions. +- [Context Optimization](https://cynrath.github.io/agent-context-kit/context-optimization/): Cut wasted agent context: duplicates, oversized files and low-value includes. +- [MCP Repository Tools](https://cynrath.github.io/agent-context-kit/mcp-repository-tools/): Expose repository context to coding agents over local MCP stdio — read-only. ## Source and package diff --git a/agent-context-kit/mcp-repository-tools/index.html b/agent-context-kit/mcp-repository-tools/index.html new file mode 100644 index 0000000..2679d89 --- /dev/null +++ b/agent-context-kit/mcp-repository-tools/index.html @@ -0,0 +1,66 @@ + + + + + + + + + MCP Repository Tools — ACKit 0.5.4 + + + + + + + + + + + + + + +

+

MCP Repository Tools

Coding agents that speak the Model Context Protocol need repository context: readiness, findings, instruction graphs, tasks and policy. Pointing them at raw files loses precedence, provenance and redaction.

ACKit serves the same deterministic analysis over local stdio MCP — 16 read-only tools including the canonical status snapshot — with no network transport and no write paths.

{ "mcpServers": { "ackit": { "command": "ackit", "args": ["mcp", "serve"] } } }

Try it without installing: npx --yes @cynrath/agent-context-kit@latest readiness

MCPProvider SurfacesGitHubnpm

+
+ + + diff --git a/agent-context-kit/mcp/index.html b/agent-context-kit/mcp/index.html index 88a5038..456e309 100644 --- a/agent-context-kit/mcp/index.html +++ b/agent-context-kit/mcp/index.html @@ -6,7 +6,7 @@ - MCP — ACKit 0.5.2 + MCP — ACKit 0.5.4 @@ -16,7 +16,7 @@ - + @@ -52,11 +52,15 @@ VS Code Security Benchmarks - Migration + Migration + AGENTS.md Validator + AI Agent Readiness + Context Optimization + MCP Repository Tools GitHub

MCP

ACKit exposes a local stdio MCP server for read-oriented repository inspection (16 read-only tools including canonical ackit_status). No remote transport is required.

{ "mcpServers": { "ackit": { "command": "ackit", "args": ["mcp", "serve"] } } }
-
+
diff --git a/agent-context-kit/migration/index.html b/agent-context-kit/migration/index.html index 02c9522..d9be7f3 100644 --- a/agent-context-kit/migration/index.html +++ b/agent-context-kit/migration/index.html @@ -6,7 +6,7 @@ - Migration — ACKit 0.5.2 + Migration — ACKit 0.5.4 @@ -16,7 +16,7 @@ - + @@ -52,13 +52,17 @@ VS Code Security Benchmarks - Migration + Migration + AGENTS.md Validator + AI Agent Readiness + Context Optimization + MCP Repository Tools GitHub -

Migration

Upgrade the package, GitHub Action and VS Code extension together so repository tooling stays on the same release line.

npm install --global @cynrath/agent-context-kit@0.5.2
+  

Migration

Upgrade the package, GitHub Action and VS Code extension together so repository tooling stays on the same release line.

npm install --global @cynrath/agent-context-kit@0.5.4
 ackit --version
-ackit sync --check
  • GitHub Action: Cynrath/agent-context-kit@v0.5.2
  • VS Code: Cynrath.ackit-vscode
  • Run ackit config check after upgrading, then ackit sync --check to reconcile ACKit-owned managed assets.
-
+ackit sync --check
  • GitHub Action: Cynrath/agent-context-kit@v0.5.4
  • VS Code: Cynrath.ackit-vscode
  • Run ackit config check after upgrading, then ackit sync --check to reconcile ACKit-owned managed assets.
+
diff --git a/agent-context-kit/optimize/index.html b/agent-context-kit/optimize/index.html index 7ca60bf..353ba29 100644 --- a/agent-context-kit/optimize/index.html +++ b/agent-context-kit/optimize/index.html @@ -6,7 +6,7 @@ - Optimize — ACKit 0.5.2 + Optimize — ACKit 0.5.4 @@ -16,7 +16,7 @@ - + @@ -52,13 +52,17 @@ VS Code Security Benchmarks - Migration + Migration + AGENTS.md Validator + AI Agent Readiness + Context Optimization + MCP Repository Tools GitHub

Optimize v2

Optimize finds redundant, conflicting or expensive repository instructions and produces evidence-backed recommendations.

ackit optimize --explain
 ackit optimize --category instruction --min-severity medium --format json
 ackit optimize --fix --dry-run
-
+
diff --git a/agent-context-kit/policy/index.html b/agent-context-kit/policy/index.html index 2048aed..08ee799 100644 --- a/agent-context-kit/policy/index.html +++ b/agent-context-kit/policy/index.html @@ -6,7 +6,7 @@ - Policy — ACKit 0.5.2 + Policy — ACKit 0.5.4 @@ -16,7 +16,7 @@ - + @@ -52,11 +52,15 @@ VS Code Security Benchmarks - Migration + Migration + AGENTS.md Validator + AI Agent Readiness + Context Optimization + MCP Repository Tools GitHub

Policy

Policy v2 combines risk-tiered autonomy (tier0 through tier4, allow / ask / deny with deny winning across layers) with review policy on policy documents and configuration.

ackit policy check
-
+
diff --git a/agent-context-kit/profiles/index.html b/agent-context-kit/profiles/index.html index 24df6b1..611422e 100644 --- a/agent-context-kit/profiles/index.html +++ b/agent-context-kit/profiles/index.html @@ -6,7 +6,7 @@ - Provider Profiles — ACKit 0.5.2 + Provider Profiles — ACKit 0.5.4 @@ -16,7 +16,7 @@ - + @@ -52,14 +52,18 @@ VS Code Security Benchmarks - Migration + Migration + AGENTS.md Validator + AI Agent Readiness + Context Optimization + MCP Repository Tools GitHub

Provider Profiles

ACKit ships profiles for Codex, Claude, Copilot, Gemini and a generic fallback.

ackit pack --profile codex --max-tokens 50000
 ackit pack --profile claude
 ackit pack --profile copilot
 ackit pack --profile gemini
-
+
diff --git a/agent-context-kit/provider-surfaces/index.html b/agent-context-kit/provider-surfaces/index.html index 3bdb948..b7de297 100644 --- a/agent-context-kit/provider-surfaces/index.html +++ b/agent-context-kit/provider-surfaces/index.html @@ -6,7 +6,7 @@ - Provider Surfaces — ACKit 0.5.2 + Provider Surfaces — ACKit 0.5.4 @@ -16,7 +16,7 @@ - + @@ -52,11 +52,15 @@ VS Code Security Benchmarks - Migration + Migration + AGENTS.md Validator + AI Agent Readiness + Context Optimization + MCP Repository Tools GitHub

Provider Surfaces

ACKit projects one canonical workflow, evidence, verdict, and status snapshot across every integration surface. Primary-source dates and caveats live with the product docs.

  • Codex CLI — AGENTS.md-native instructions with nested scope.
  • Claude Code — CLAUDE.md-native instructions with rules paths.
  • Gemini CLI — GEMINI.md instructions with JIT discovery.
  • Copilot Chat — copilot-instructions plus AGENTS.md with applyTo scopes.
  • Copilot coding agent — multi-native AGENTS plus CLAUDE plus GEMINI.
  • Copilot review — AGENTS-only channel with repo toggle and HEAD sourcing.
  • VS Code agent mode — settings-gated multi-native instructions.

CLI (ackit status), SDK (buildStatusReport), read-only MCP (ackit_status), the GitHub Action (command: status), and the VS Code Tasks view all expose the same snapshot.

-
+
diff --git a/agent-context-kit/readiness/index.html b/agent-context-kit/readiness/index.html index c322a8e..b7b53cc 100644 --- a/agent-context-kit/readiness/index.html +++ b/agent-context-kit/readiness/index.html @@ -6,7 +6,7 @@ - Readiness — ACKit 0.5.2 + Readiness — ACKit 0.5.4 @@ -16,7 +16,7 @@ - + @@ -52,13 +52,17 @@ VS Code Security Benchmarks - Migration + Migration + AGENTS.md Validator + AI Agent Readiness + Context Optimization + MCP Repository Tools GitHub

Readiness

Readiness summarizes whether a repository exposes enough instructions, security controls, context discipline, task state, skills and policy for repeatable agent-assisted work.

ackit readiness
 ackit readiness --fail-below 80 --strict
 ackit readiness --baseline .ackit/readiness.json --compare

Score model

Six categories are weighted and renormalized when a category is not applicable. JSON output uses the stable ackit.readiness.v1 contract.

-
+
diff --git a/agent-context-kit/roles/index.html b/agent-context-kit/roles/index.html index 2078242..e881ebc 100644 --- a/agent-context-kit/roles/index.html +++ b/agent-context-kit/roles/index.html @@ -6,7 +6,7 @@ - Roles — ACKit 0.5.2 + Roles — ACKit 0.5.4 @@ -16,7 +16,7 @@ - + @@ -52,13 +52,17 @@ VS Code Security Benchmarks - Migration + Migration + AGENTS.md Validator + AI Agent Readiness + Context Optimization + MCP Repository Tools GitHub

Roles

Role contracts describe researcher, architect, implementer, verifier, security-reviewer, documentation-reviewer and release-reviewer responsibilities as portable data-only documents.

ackit role list
 ackit role show verifier
 ackit role validate
-
+
diff --git a/agent-context-kit/rule-packs/index.html b/agent-context-kit/rule-packs/index.html index 281be47..048f4ad 100644 --- a/agent-context-kit/rule-packs/index.html +++ b/agent-context-kit/rule-packs/index.html @@ -6,7 +6,7 @@ - Rule Packs — ACKit 0.5.2 + Rule Packs — ACKit 0.5.4 @@ -16,7 +16,7 @@ - + @@ -52,11 +52,15 @@ VS Code Security Benchmarks - Migration + Migration + AGENTS.md Validator + AI Agent Readiness + Context Optimization + MCP Repository Tools GitHub

Rule / Policy Packs

Rule packs turn repository requirements into machine-checkable presence, pattern, config, dependency and instruction checks.

ackit policy check
-
+
diff --git a/agent-context-kit/sdk/index.html b/agent-context-kit/sdk/index.html index 5eb323b..7710bf3 100644 --- a/agent-context-kit/sdk/index.html +++ b/agent-context-kit/sdk/index.html @@ -6,7 +6,7 @@ - SDK — ACKit 0.5.2 + SDK — ACKit 0.5.4 @@ -16,7 +16,7 @@ - + @@ -52,12 +52,16 @@ VS Code Security Benchmarks - Migration + Migration + AGENTS.md Validator + AI Agent Readiness + Context Optimization + MCP Repository Tools GitHub

SDK

The ESM SDK exposes the same deterministic repository analysis primitives used by the CLI.

import { scanRepository } from "@cynrath/agent-context-kit";
 const result = await scanRepository({ canonicalPath: process.cwd() });

Read-model parity: buildStatusReport and renderStatusReport project the same ackit.status.v1 snapshot as ackit status; handoff and independence exports share the CLI contract.

-
+
diff --git a/agent-context-kit/security/index.html b/agent-context-kit/security/index.html index 5fd5883..ec2adfe 100644 --- a/agent-context-kit/security/index.html +++ b/agent-context-kit/security/index.html @@ -6,7 +6,7 @@ - Security — ACKit 0.5.2 + Security — ACKit 0.5.4 @@ -16,7 +16,7 @@ - + @@ -52,11 +52,15 @@ VS Code Security Benchmarks - Migration + Migration + AGENTS.md Validator + AI Agent Readiness + Context Optimization + MCP Repository Tools GitHub

Security

Offline-first: product analysis is designed for zero outbound repository-data egress after installation.

  • No telemetry or hosted LLM calls in repository analysis.
  • Evidence is redacted at construction.
  • Path containment and traversal checks protect file access.
  • Rule evaluation includes ReDoS and size guards.
  • Dashboard defaults to loopback with CSP.
  • v0.5 containment delta: link-aware output guard with symlink/junction escape audit and proven-gap-only repair; MCP remains read-only.
-
+
diff --git a/agent-context-kit/skills/index.html b/agent-context-kit/skills/index.html index 3f74196..95be8d4 100644 --- a/agent-context-kit/skills/index.html +++ b/agent-context-kit/skills/index.html @@ -6,7 +6,7 @@ - Skills — ACKit 0.5.2 + Skills — ACKit 0.5.4 @@ -16,7 +16,7 @@ - + @@ -52,7 +52,11 @@ VS Code Security Benchmarks - Migration + Migration + AGENTS.md Validator + AI Agent Readiness + Context Optimization + MCP Repository Tools GitHub

Skills

ACKit manages open-standard skills with lock-tracked installs, ownership and conflict handling, plus deterministic projections for Claude Code, Copilot and generic layouts.

ackit skills list
@@ -60,7 +64,7 @@
 ackit skills install
 ackit skills sync
 ackit skills export --provider claude --out ./skills-out
-
+
diff --git a/agent-context-kit/status/index.html b/agent-context-kit/status/index.html index 2fad0a2..78b062b 100644 --- a/agent-context-kit/status/index.html +++ b/agent-context-kit/status/index.html @@ -6,7 +6,7 @@ - Status — ACKit 0.5.2 + Status — ACKit 0.5.4 @@ -16,7 +16,7 @@ - + @@ -52,13 +52,17 @@ VS Code Security Benchmarks - Migration + Migration + AGENTS.md Validator + AI Agent Readiness + Context Optimization + MCP Repository Tools GitHub

Status

ackit status [id] is the canonical read-only projection: task and stage, verbatim completion blockers, verification and checkpoint freshness, and derived next actions. Human output and stable ackit.status.v1 JSON describe the same snapshot; the command never mutates.

ackit status
 ackit status TASK-0007
 ackit --json status TASK-0007
-
+
diff --git a/agent-context-kit/sync/index.html b/agent-context-kit/sync/index.html index f4d2c6e..664d842 100644 --- a/agent-context-kit/sync/index.html +++ b/agent-context-kit/sync/index.html @@ -6,7 +6,7 @@ - Managed Sync — ACKit 0.5.2 + Managed Sync — ACKit 0.5.4 @@ -16,7 +16,7 @@ - + @@ -52,14 +52,18 @@ VS Code Security Benchmarks - Migration + Migration + AGENTS.md Validator + AI Agent Readiness + Context Optimization + MCP Repository Tools GitHub

Managed Asset Sync

ackit sync reconciles ACKit-owned managed assets (instruction blocks and builtin skills) after upgrades. Write decisions are content-driven: upgrading the package alone never rewrites files.

ackit sync --dry-run
 ackit sync --check
 ackit sync
 ackit sync --force
-
+
diff --git a/agent-context-kit/trust-flow/index.html b/agent-context-kit/trust-flow/index.html index 68602b7..dfb7d87 100644 --- a/agent-context-kit/trust-flow/index.html +++ b/agent-context-kit/trust-flow/index.html @@ -6,7 +6,7 @@ - Trust Flow — ACKit 0.5.2 + Trust Flow — ACKit 0.5.4 @@ -16,7 +16,7 @@ - + @@ -52,7 +52,11 @@ VS Code Security Benchmarks - Migration + Migration + AGENTS.md Validator + AI Agent Readiness + Context Optimization + MCP Repository Tools GitHub

Trust Flow

One reproducible offline arc: missing evidence and verdicts block completion, a bound independent verdict enables it, a state mutation stales it, fresh re-verification restores it, a checkpoint handoff exports it, and a second process resumes it with zero shared memory.

ackit task complete TASK-0007
@@ -61,7 +65,7 @@
 ackit status TASK-0007
 ackit checkpoint export TASK-0007 --format json --out .ackit/reviews/handoff.json
 ackit checkpoint import .ackit/reviews/handoff.json

Executed by tests/e2e/trust-flow-demo.test.ts and walked through in the product guide docs/guides/demo-trust-flow.md.

-
+
diff --git a/agent-context-kit/verification/index.html b/agent-context-kit/verification/index.html index dfbe369..1bdd202 100644 --- a/agent-context-kit/verification/index.html +++ b/agent-context-kit/verification/index.html @@ -6,7 +6,7 @@ - Verification — ACKit 0.5.2 + Verification — ACKit 0.5.4 @@ -16,7 +16,7 @@ - + @@ -52,13 +52,17 @@ VS Code Security Benchmarks - Migration + Migration + AGENTS.md Validator + AI Agent Readiness + Context Optimization + MCP Repository Tools GitHub

Verification / Verdicts

Verification bundles (ackit.verification-bundle.v2) give a fresh independent verifier everything needed to judge a task, bound to the exact state via stateDigest and bundleDigest. The verifier authors ackit.verdict.v1; ACKit persists a state-bound ackit.verdict.v2. Legacy v1 files stay readable as history but never satisfy a state-bound completion.

ackit verification bundle --format json --out .ackit/reviews/bundle.json
 ackit verification record --verdict ./verdict.yaml --bundle .ackit/reviews/bundle.json
 ackit verification show

Independence and freshness

Independence is proven by reviewedBundleDigest: fresh claims without bundle proof refuse with VERDICT-INDEPENDENCE-UNPROVEN, replayed content refuses with VERDICT-REPLAY-REJECTED, stale bundles with VERDICT-BUNDLE-MISMATCH, and moved state with VERDICT-STATE-STALE. Completion rechecks freshness; no cryptographic verifier identity is claimed.

-
+
diff --git a/agent-context-kit/vscode/index.html b/agent-context-kit/vscode/index.html index d511d7b..ec3d230 100644 --- a/agent-context-kit/vscode/index.html +++ b/agent-context-kit/vscode/index.html @@ -6,7 +6,7 @@ - VS Code — ACKit 0.5.2 + VS Code — ACKit 0.5.4 @@ -16,7 +16,7 @@ - + @@ -52,11 +52,15 @@ VS Code Security Benchmarks - Migration + Migration + AGENTS.md Validator + AI Agent Readiness + Context Optimization + MCP Repository Tools GitHub

VS Code

Cynrath.ackit-vscode surfaces readiness, findings, instructions, tasks, policy and diagnostics inside VS Code. The Tasks view renders the canonical status snapshot (blockers plus next actions) with no write paths added.

code --install-extension Cynrath.ackit-vscode
-
+
diff --git a/agent-context-kit/workflows/index.html b/agent-context-kit/workflows/index.html index e0aae1d..e2b858a 100644 --- a/agent-context-kit/workflows/index.html +++ b/agent-context-kit/workflows/index.html @@ -6,7 +6,7 @@ - Workflows — ACKit 0.5.2 + Workflows — ACKit 0.5.4 @@ -16,7 +16,7 @@ - + @@ -52,14 +52,18 @@ VS Code Security Benchmarks - Migration + Migration + AGENTS.md Validator + AI Agent Readiness + Context Optimization + MCP Repository Tools GitHub

Workflows

Workflows run repository work through staged lifecycles (quick, standard, high-risk) with required artifacts and deterministic gates. Optional workflow: keys in ackit.yml tune the built-in profile minimums; repositories without them keep exact prior behavior.

ackit workflow set --profile standard
 ackit workflow show
 ackit workflow advance
 ackit workflow verify --outcome pass
-
+
diff --git a/index.html b/index.html index db9557a..8423c4f 100644 --- a/index.html +++ b/index.html @@ -1 +1 @@ -Cynrath — Developer tooling, web systems & infrastructure
Developer tooling · web systems · infrastructure

Developer tools.
Web systems.
Infrastructure.

TypeScript and .NET for CLI tooling, admin platforms, databases, automation, CI/CD and infrastructure.

TypeScript / Node.js.NET / C#Docker / LinuxCloudflare / CI
ackit / localv0.5.2
$ackit readiness
Readiness88/100██████████████████░░threshold 80 · pass✓ Instructions 90/100✓ Security 90/100✓ Context Efficiency 70/100
$
14stars
1fork
0open issues
Stack
TypeScriptNode.js.NET / C#MSSQL / EF CoreDocker / LinuxCloudflare
02 / AgentContextKit

Repository context you can inspect.

ACKit runs locally and turns repository instructions, context, tasks, policies, workflow state, evidence, verification, diagnostics and integrations into deterministic commands and stable outputs.

Installnpm install -g @cynrath/agent-context-kit@0.5.2
01

Agent Readiness

Explainable 0–100 scoring across repository readiness categories with strict, baseline and compare modes.

ackit readiness
02

Instruction Graph v2

Resolve Codex, Claude, Gemini, Copilot and shared instructions with scope, precedence, provenance and conflict analysis.

ackit instructions --explain
03

Provider Profiles

Deterministic provider-aware behavior for Codex, Claude, Copilot, Gemini or generic workflows; CLI, SDK, MCP, Action and VS Code share one status snapshot.

ackit pack --profile codex
04

Context Packs

Build deterministic, token-budgeted repository context with weighted ranking and per-file manifest reasons.

ackit pack --max-tokens 50000
05

Skills Interop

Validate, install and export portable skills across supported provider layouts without replacing the repository workflow.

ackit skills export --provider claude
06

Managed Asset Sync

Preview, check and reconcile ACKit-owned AGENTS blocks, provider shims and builtin skills without overwriting user-owned content.

ackit sync --dry-run
07

Rule + Policy Packs

Apply machine-checkable presence, pattern, config, dependency and instruction rules with composition and scoped overrides.

ackit policy check
08

Optimize v2

Detect instruction waste, duplicates and conflicts with evidence, confidence, token estimates and dry-run fixes.

ackit optimize --explain
09

Security Scanning

Scan secrets, keys, connection strings, entropy, CI pinning and hygiene findings with redacted evidence and SARIF; link-aware containment hardens outputs.

ackit scan --ci
10

Task System

Keep a single active task, explicit task states, artifact references, completion gates and doctor checks.

ackit task
11

Workflows + Intent

Run quick, standard or high-risk stage lifecycles backed by committed intent, plans and config-driven completion gates.

ackit workflow show
12

Evidence + State-Bound Verification

Link criteria to typed evidence, verify against state-bound bundles and persist independent bound verdicts with replay and staleness refusal.

ackit verification record --bundle
13

Portable Handoff + Canonical Status

Resume anywhere with verification-bound handoff packs and the read-only status snapshot: blockers, freshness and next actions.

ackit checkpoint import
14

Drift + Policy v2

Detect deterministic workflow drift and enforce risk-tiered autonomy, review policy and declarative lifecycle gates.

ackit drift check --ci
15

Roles + Journal

Use portable role contracts and a sanitized append-only local execution journal without capturing hidden reasoning or conversations.

ackit role list
16

Watch + Dashboard

Watch repository changes and inspect local readiness, findings and workflow state through a localhost-only dashboard.

ackit dashboard
17

Diagnostics + Benchmarks

Produce deterministic diagnostic bundles and measure scan, pack, graph, cache and memory performance against fixtures.

ackit diagnostics bundle
18

SDK + MCP

Embed ACKit through its typed SDK or expose the read-only MCP surface for agent integrations.

ackit mcp serve
19

VS Code + GitHub Action

Bring ACKit checks into editor Problems, the status-integrated Tasks view and CI with the VS Code extension and GitHub Action.

Cynrath.ackit-vscode
20

Trust Flow Demo

Follow the reproducible offline arc from evidence to bound independent verification, status and second-process resume — no API key.

ackit status
04 / Engineering principles

The rules behind the work.

01

Read before edit.

Architecture, instructions, local changes and task state come first.

Inspect AGENTS/CLAUDE files, docs, ADRs, git status/diff and the repository's real build/test workflow.
02

Plan before implementation.

Tasks, acceptance criteria and verification are written down before code starts.

The plan stays usable across long sessions, interruptions and handoffs.
03

Smallest safe change.

Fix the root cause without dragging unrelated architecture into the change.

Preserve conventions, dependencies and unrelated local work unless they are part of the actual problem.
04

Verify before done.

Builds, tests, E2E, CI and the final diff decide completion.

A change is complete only when the required checks pass and the final diff is clean.
\ No newline at end of file +Cynrath — Developer tooling, web systems & infrastructure
Developer tooling · web systems · infrastructure

Developer tools.
Web systems.
Infrastructure.

TypeScript and .NET for CLI tooling, admin platforms, databases, automation, CI/CD and infrastructure.

TypeScript / Node.js.NET / C#Docker / LinuxCloudflare / CI
ackit / localv0.5.4
$ackit readiness
Readiness88/100██████████████████░░threshold 80 · pass✓ Instructions 90/100✓ Security 90/100✓ Context Efficiency 70/100
$
14stars
1fork
0open issues
Stack
TypeScriptNode.js.NET / C#MSSQL / EF CoreDocker / LinuxCloudflare
02 / AgentContextKit

Repository context you can inspect.

ACKit runs locally and turns repository instructions, context, tasks, policies, workflow state, evidence, verification, diagnostics and integrations into deterministic commands and stable outputs.

Installnpm install -g @cynrath/agent-context-kit@0.5.4
01

Agent Readiness

Explainable 0–100 scoring across repository readiness categories with strict, baseline and compare modes.

ackit readiness
02

Instruction Graph v2

Resolve Codex, Claude, Gemini, Copilot and shared instructions with scope, precedence, provenance and conflict analysis.

ackit instructions --explain
03

Provider Profiles

Deterministic provider-aware behavior for Codex, Claude, Copilot, Gemini or generic workflows; CLI, SDK, MCP, Action and VS Code share one status snapshot.

ackit pack --profile codex
04

Context Packs

Build deterministic, token-budgeted repository context with weighted ranking and per-file manifest reasons.

ackit pack --max-tokens 50000
05

Skills Interop

Validate, install and export portable skills across supported provider layouts without replacing the repository workflow.

ackit skills export --provider claude
06

Managed Asset Sync

Preview, check and reconcile ACKit-owned AGENTS blocks, provider shims and builtin skills without overwriting user-owned content.

ackit sync --dry-run
07

Rule + Policy Packs

Apply machine-checkable presence, pattern, config, dependency and instruction rules with composition and scoped overrides.

ackit policy check
08

Optimize v2

Detect instruction waste, duplicates and conflicts with evidence, confidence, token estimates and dry-run fixes.

ackit optimize --explain
09

Security Scanning

Scan secrets, keys, connection strings, entropy, CI pinning and hygiene findings with redacted evidence and SARIF; link-aware containment hardens outputs.

ackit scan --ci
10

Task System

Keep a single active task, explicit task states, artifact references, completion gates and doctor checks.

ackit task
11

Workflows + Intent

Run quick, standard or high-risk stage lifecycles backed by committed intent, plans and config-driven completion gates.

ackit workflow show
12

Evidence + State-Bound Verification

Link criteria to typed evidence, verify against state-bound bundles and persist independent bound verdicts with replay and staleness refusal.

ackit verification record --bundle
13

Portable Handoff + Canonical Status

Resume anywhere with verification-bound handoff packs and the read-only status snapshot: blockers, freshness and next actions.

ackit checkpoint import
14

Drift + Policy v2

Detect deterministic workflow drift and enforce risk-tiered autonomy, review policy and declarative lifecycle gates.

ackit drift check --ci
15

Roles + Journal

Use portable role contracts and a sanitized append-only local execution journal without capturing hidden reasoning or conversations.

ackit role list
16

Watch + Dashboard

Watch repository changes and inspect local readiness, findings and workflow state through a localhost-only dashboard.

ackit dashboard
17

Diagnostics + Benchmarks

Produce deterministic diagnostic bundles and measure scan, pack, graph, cache and memory performance against fixtures.

ackit diagnostics bundle
18

SDK + MCP

Embed ACKit through its typed SDK or expose the read-only MCP surface for agent integrations.

ackit mcp serve
19

VS Code + GitHub Action

Bring ACKit checks into editor Problems, the status-integrated Tasks view and CI with the VS Code extension and GitHub Action.

Cynrath.ackit-vscode
20

Trust Flow Demo

Follow the reproducible offline arc from evidence to bound independent verification, status and second-process resume — no API key.

ackit status
04 / Engineering principles

The rules behind the work.

01

Read before edit.

Architecture, instructions, local changes and task state come first.

Inspect AGENTS/CLAUDE files, docs, ADRs, git status/diff and the repository's real build/test workflow.
02

Plan before implementation.

Tasks, acceptance criteria and verification are written down before code starts.

The plan stays usable across long sessions, interruptions and handoffs.
03

Smallest safe change.

Fix the root cause without dragging unrelated architecture into the change.

Preserve conventions, dependencies and unrelated local work unless they are part of the actual problem.
04

Verify before done.

Builds, tests, E2E, CI and the final diff decide completion.

A change is complete only when the required checks pass and the final diff is clean.
\ No newline at end of file diff --git a/scripts/sync-ackit-docs.mjs b/scripts/sync-ackit-docs.mjs index 5e4995c..9c11dd0 100644 --- a/scripts/sync-ackit-docs.mjs +++ b/scripts/sync-ackit-docs.mjs @@ -80,7 +80,9 @@ const nav = [ ['sync','Managed Sync'],['github-action','GitHub Action'],['mcp','MCP'],['sdk','SDK'], ['provider-surfaces','Provider Surfaces'],['trust-flow','Trust Flow'], ['dashboard','Dashboard'],['diagnostics','Diagnostics'],['vscode','VS Code'],['security','Security'], - ['benchmarks','Benchmarks'],['migration','Migration'] + ['benchmarks','Benchmarks'],['migration','Migration'], + ['agents-md-validator','AGENTS.md Validator'],['ai-agent-readiness','AI Agent Readiness'], + ['context-optimization','Context Optimization'],['mcp-repository-tools','MCP Repository Tools'] ]; function template({ title, description, slug, version, body }) { @@ -132,7 +134,7 @@ function template({ title, description, slug, version, body }) { function pages(version) { const install = `npm install --global @cynrath/agent-context-kit@${version}\nackit --version\nackit --help`; return [ - {slug:'',title:'AgentContextKit',description:'Offline-first deterministic toolkit for agent-ready repositories.',body:`

AgentContextKit ${esc(version)}

Turn repository context into something agents and humans can inspect. ACKit combines readiness scoring, instruction graphs, context packs, policy-as-code, security scanning, tasks, MCP, dashboard and editor tooling without sending repository data to a hosted analysis service.

Getting StartedCLI ReferenceGitHubnpm

Install

${esc(install)}

Quickstart

ackit init --dry-run\nackit readiness\nackit instructions --explain\nackit scan --ci\nackit sync --check\nackit pack --profile codex --max-tokens 50000\nackit dashboard --port 0 --open

Core capabilities

  • Agent Readiness — explainable 0–100 scoring with strict gates and baselines.
  • Instruction Graph v2 — provider-aware scope, precedence, provenance, conflicts and duplicates.
  • Context Packs — deterministic ranking, token budgets and per-file manifests.
  • Policy Packs — machine-checkable repository rules.
  • Scanning + SARIF — secrets, connection strings, entropy, absolute paths, CI pinning and redacted evidence.
  • Tasks — docs-first single-active workflow and completion gates.
  • Workflows + Intent — staged lifecycles with gates, committed intent docs with fingerprints.
  • Evidence + State-Bound Verification — criteria-to-proof registry, state-bound bundles (ackit.verification-bundle.v2), persisted bound verdicts (ackit.verdict.v2) with verifier independence and replay/staleness refusal.
  • Checkpoints + Portable Handoff v2 — snapshots, verification-bound portable handoff (ackit.handoff.v2), read-only import with stale refusal.
  • Canonical Status — read-only task/stage, blockers, verification and checkpoint freshness, next actions (ackit.status.v1).
  • Provider / Surface Parity — CLI, SDK, read-only MCP, Action and VS Code project the same status snapshot.
  • Containment Hardening — link-aware output guard with symlink/junction escape repair; MCP stays read-only.
  • Trust-Flow Demo — reproducible offline proof from missing evidence to second-process resume.
  • Drift + Policy v2 — deterministic drift findings, risk-tiered autonomy and review policy.
  • Roles + Skills — portable role contracts, managed open-standard skills.
  • Managed Sync — content-driven reconciliation of ACKit-owned assets after upgrades.
  • MCP + SDK — local integrations for agents and tooling.
  • Dashboard + VS Code — local inspection surfaces for readiness, findings, graph, tasks and policy.
`}, + {slug:'',title:'AgentContextKit',description:'Agent readiness and repository context toolkit for Codex, Claude Code, Cursor, Copilot, Gemini and MCP-based coding agents. Offline-first, deterministic.',body:`

AgentContextKit ${esc(version)}

Turn repository context into something agents and humans can inspect. ACKit combines readiness scoring, instruction graphs, context packs, policy-as-code, security scanning, tasks, MCP, dashboard and editor tooling without sending repository data to a hosted analysis service.

Agent context and repository readiness for Codex, Claude Code, Cursor, Copilot and Gemini — validate agent instructions, scan repositories, optimize context and enforce evidence-backed workflows. Offline-first, deterministic, no telemetry.

Getting StartedCLI ReferenceGitHubnpm

Try it without installing: npx --yes @cynrath/agent-context-kit@latest readiness — one command scores any repository 0–100 with actionable findings.

Install

${esc(install)}

Quickstart

ackit init --dry-run\nackit readiness\nackit instructions --explain\nackit scan --ci\nackit sync --check\nackit pack --profile codex --max-tokens 50000\nackit dashboard --port 0 --open

Core capabilities

  • Agent Readiness — explainable 0–100 scoring with strict gates and baselines.
  • Instruction Graph v2 — provider-aware scope, precedence, provenance, conflicts and duplicates.
  • Context Packs — deterministic ranking, token budgets and per-file manifests.
  • Policy Packs — machine-checkable repository rules.
  • Scanning + SARIF — secrets, connection strings, entropy, absolute paths, CI pinning and redacted evidence.
  • Tasks — docs-first single-active workflow and completion gates.
  • Workflows + Intent — staged lifecycles with gates, committed intent docs with fingerprints.
  • Evidence + State-Bound Verification — criteria-to-proof registry, state-bound bundles (ackit.verification-bundle.v2), persisted bound verdicts (ackit.verdict.v2) with verifier independence and replay/staleness refusal.
  • Checkpoints + Portable Handoff v2 — snapshots, verification-bound portable handoff (ackit.handoff.v2), read-only import with stale refusal.
  • Canonical Status — read-only task/stage, blockers, verification and checkpoint freshness, next actions (ackit.status.v1).
  • Provider / Surface Parity — CLI, SDK, read-only MCP, Action and VS Code project the same status snapshot.
  • Containment Hardening — link-aware output guard with symlink/junction escape repair; MCP stays read-only.
  • Trust-Flow Demo — reproducible offline proof from missing evidence to second-process resume.
  • Drift + Policy v2 — deterministic drift findings, risk-tiered autonomy and review policy.
  • Roles + Skills — portable role contracts, managed open-standard skills.
  • Managed Sync — content-driven reconciliation of ACKit-owned assets after upgrades.
  • MCP + SDK — local integrations for agents and tooling.
  • Dashboard + VS Code — local inspection surfaces for readiness, findings, graph, tasks and policy.
`}, {slug:'getting-started',title:'Getting Started',description:'Install and initialize AgentContextKit.',body:`

Getting Started

ACKit requires Node.js 22 or newer. Installation is global or via npx; repository analysis runs locally.

Install

${esc(install)}

First pass

ackit init --dry-run\nackit readiness\nackit scan --ci\nackit instructions --explain
`}, {slug:'cli',title:'CLI Reference',description:'ACKit commands, options and exit behavior.',body:`

CLI Reference

The CLI favors stable terminal and JSON output so the same checks can be used interactively and in CI.

ackit init --dry-run\nackit sync --check\nackit readiness --fail-below 80 --strict --baseline --compare --json\nackit scan --ci --changed --staged --since --range --baseline --watch\nackit instructions --provider codex --profile codex --for <path> --explain --json\nackit pack --profile codex --max-tokens 50000 --include <glob> --changed\nackit optimize --explain --fix --dry-run --format json\nackit policy check\nackit task create | list | start | complete\nackit workflow set | show | advance | verify\nackit intent new | list | show | validate | fingerprint\nackit checkpoint create | show | validate | export --format json\nackit checkpoint import <file>\nackit status [id]\nackit evidence sync | show | verify | validate\nackit verification bundle --format json | record --bundle | show\nackit drift check --ci\nackit role list | show | validate\nackit skills list | validate | install | sync\nackit journal show | validate\nackit diagnostics --json\nackit diagnostics bundle --out ./ackit-diag.zip --redact-check\nackit dashboard --host 127.0.0.1 --port 0 --open\nackit mcp serve

Exit codes

0 success, 1 threshold failure, 2 usage/configuration, 3 environment, 4 security, 5 internal error.

`}, {slug:'readiness',title:'Readiness',description:'Explainable ACKit repository readiness scoring.',body:`

Readiness

Readiness summarizes whether a repository exposes enough instructions, security controls, context discipline, task state, skills and policy for repeatable agent-assisted work.

ackit readiness\nackit readiness --fail-below 80 --strict\nackit readiness --baseline .ackit/readiness.json --compare

Score model

Six categories are weighted and renormalized when a category is not applicable. JSON output uses the stable ackit.readiness.v1 contract.

`}, @@ -161,7 +163,11 @@ function pages(version) { {slug:'vscode',title:'VS Code',description:'AgentContextKit VS Code extension.',body:`

VS Code

Cynrath.ackit-vscode surfaces readiness, findings, instructions, tasks, policy and diagnostics inside VS Code. The Tasks view renders the canonical status snapshot (blockers plus next actions) with no write paths added.

code --install-extension Cynrath.ackit-vscode
`}, {slug:'security',title:'Security',description:'ACKit offline-first threat model and redaction guarantees.',body:`

Security

Offline-first: product analysis is designed for zero outbound repository-data egress after installation.

  • No telemetry or hosted LLM calls in repository analysis.
  • Evidence is redacted at construction.
  • Path containment and traversal checks protect file access.
  • Rule evaluation includes ReDoS and size guards.
  • Dashboard defaults to loopback with CSP.
  • v0.5 containment delta: link-aware output guard with symlink/junction escape audit and proven-gap-only repair; MCP remains read-only.
`}, {slug:'benchmarks',title:'Benchmarks',description:'Deterministic ACKit performance benchmarks.',body:`

Benchmarks

ACKit benchmarks cold scan, warm scan, incremental scan, memory, throughput, context-pack construction, instruction-graph construction and cache behavior.

node benchmarks/run.mjs --classes small --out /tmp/out\nnode benchmarks/check-thresholds.mjs
`}, - {slug:'migration',title:'Migration',description:'Upgrade AgentContextKit to the current release.',body:`

Migration

Upgrade the package, GitHub Action and VS Code extension together so repository tooling stays on the same release line.

npm install --global @cynrath/agent-context-kit@${esc(version)}\nackit --version\nackit sync --check
  • GitHub Action: Cynrath/agent-context-kit@v${esc(version)}
  • VS Code: Cynrath.ackit-vscode
  • Run ackit config check after upgrading, then ackit sync --check to reconcile ACKit-owned managed assets.
`} + {slug:'migration',title:'Migration',description:'Upgrade AgentContextKit to the current release.',body:`

Migration

Upgrade the package, GitHub Action and VS Code extension together so repository tooling stays on the same release line.

npm install --global @cynrath/agent-context-kit@${esc(version)}\nackit --version\nackit sync --check
  • GitHub Action: Cynrath/agent-context-kit@v${esc(version)}
  • VS Code: Cynrath.ackit-vscode
  • Run ackit config check after upgrading, then ackit sync --check to reconcile ACKit-owned managed assets.
`}, + {slug:'agents-md-validator',title:'AGENTS.md Validator',description:'Validate AGENTS.md and CLAUDE.md instructions; find conflicts, duplicates and scope problems.',body:`

AGENTS.md Validator

One repository, many agents: AGENTS.md (Codex), CLAUDE.md (Claude Code), GEMINI.md (Gemini) and Copilot instructions plus nested scopes. When files disagree, agents follow stale or contradictory guidance and waste runs.

ACKit resolves every instruction file into a provider-aware graph with scope, precedence and provenance, then reports conflicts, duplicates, shadowing and dead files — before your agent reads them.

ackit instructions --explain\nackit optimize --explain
codex/AGENTS.md → claude/CLAUDE.md (provenance: frontmatter applyTo)

Try it without installing: npx --yes @cynrath/agent-context-kit@latest readiness

Instruction GraphOptimizeGitHubnpm

`}, + {slug:'ai-agent-readiness',title:'AI Agent Readiness',description:'Score any repository 0–100 for AI coding-agent readiness with explainable deductions.',body:`

AI Agent Readiness

Is your repository actually ready for Codex, Claude Code, Cursor, Copilot or Gemini — or will the agent burn context on missing instructions, leaked secrets and bloated includes? Guessing costs tokens; measuring is one command.

ACKit scores instructions, security, context efficiency, task hygiene, skills and policy on a deterministic 0–100 scale with weighted renormalization, strict gates and baselines.

ackit readiness\nackit readiness --strict\nackit readiness --fail-below 80
Readiness  89/100  ██████████████████░░  (threshold 0 — pass)\n  Instructions          90/100\n  Security              90/100\n  Context Efficiency    70/100

Try it without installing: npx --yes @cynrath/agent-context-kit@latest readiness

ReadinessCLI ReferenceGitHubnpm

`}, + {slug:'context-optimization',title:'Context Optimization',description:'Cut wasted agent context: duplicates, oversized files and low-value includes.',body:`

Context Optimization

Every duplicated guidance block and oversized context file is tokens your agent pays for on every run. ACKit finds the waste with evidence — file paths, token-waste estimates and provenance — and proposes safe, reviewable fixes.

Pair the advisor with budgeted, provider-aware context packs so each agent gets exactly the context it needs.

ackit optimize --explain\nackit pack --profile codex --max-tokens 50000
# Optimize Report (1 suggestion(s))\n- **[low] budget-overrun** (low): 867 file(s) did not fit the 20000-token budget

Try it without installing: npx --yes @cynrath/agent-context-kit@latest readiness

OptimizeProvider ProfilesGitHubnpm

`}, + {slug:'mcp-repository-tools',title:'MCP Repository Tools',description:'Expose repository context to coding agents over local MCP stdio — read-only.',body:`

MCP Repository Tools

Coding agents that speak the Model Context Protocol need repository context: readiness, findings, instruction graphs, tasks and policy. Pointing them at raw files loses precedence, provenance and redaction.

ACKit serves the same deterministic analysis over local stdio MCP — 16 read-only tools including the canonical status snapshot — with no network transport and no write paths.

{ "mcpServers": { "ackit": { "command": "ackit", "args": ["mcp", "serve"] } } }

Try it without installing: npx --yes @cynrath/agent-context-kit@latest readiness

MCPProvider SurfacesGitHubnpm

`} ]; } diff --git a/sitemap.xml b/sitemap.xml index 4cbb9cd..18bb200 100644 --- a/sitemap.xml +++ b/sitemap.xml @@ -2,126 +2,142 @@ https://cynrath.github.io/ - 2026-09-06 + 2026-09-22 https://cynrath.github.io/agent-context-kit/ - 2026-09-06 + 2026-09-22 + + + https://cynrath.github.io/agent-context-kit/agents-md-validator/ + 2026-09-22 + + + https://cynrath.github.io/agent-context-kit/ai-agent-readiness/ + 2026-09-22 https://cynrath.github.io/agent-context-kit/benchmarks/ - 2026-09-06 + 2026-09-22 https://cynrath.github.io/agent-context-kit/checkpoints/ - 2026-09-06 + 2026-09-22 https://cynrath.github.io/agent-context-kit/cli/ - 2026-09-06 + 2026-09-22 + + + https://cynrath.github.io/agent-context-kit/context-optimization/ + 2026-09-22 https://cynrath.github.io/agent-context-kit/dashboard/ - 2026-09-06 + 2026-09-22 https://cynrath.github.io/agent-context-kit/diagnostics/ - 2026-09-06 + 2026-09-22 https://cynrath.github.io/agent-context-kit/drift/ - 2026-09-06 + 2026-09-22 https://cynrath.github.io/agent-context-kit/evidence/ - 2026-09-06 + 2026-09-22 https://cynrath.github.io/agent-context-kit/getting-started/ - 2026-09-06 + 2026-09-22 https://cynrath.github.io/agent-context-kit/github-action/ - 2026-09-06 + 2026-09-22 https://cynrath.github.io/agent-context-kit/instruction-graph/ - 2026-09-06 + 2026-09-22 https://cynrath.github.io/agent-context-kit/intent/ - 2026-09-06 + 2026-09-22 + + + https://cynrath.github.io/agent-context-kit/mcp-repository-tools/ + 2026-09-22 https://cynrath.github.io/agent-context-kit/mcp/ - 2026-09-06 + 2026-09-22 https://cynrath.github.io/agent-context-kit/migration/ - 2026-09-06 + 2026-09-22 https://cynrath.github.io/agent-context-kit/optimize/ - 2026-09-06 + 2026-09-22 https://cynrath.github.io/agent-context-kit/policy/ - 2026-09-06 + 2026-09-22 https://cynrath.github.io/agent-context-kit/profiles/ - 2026-09-06 + 2026-09-22 https://cynrath.github.io/agent-context-kit/provider-surfaces/ - 2026-09-06 + 2026-09-22 https://cynrath.github.io/agent-context-kit/readiness/ - 2026-09-06 + 2026-09-22 https://cynrath.github.io/agent-context-kit/roles/ - 2026-09-06 + 2026-09-22 https://cynrath.github.io/agent-context-kit/rule-packs/ - 2026-09-06 + 2026-09-22 https://cynrath.github.io/agent-context-kit/sdk/ - 2026-09-06 + 2026-09-22 https://cynrath.github.io/agent-context-kit/security/ - 2026-09-06 + 2026-09-22 https://cynrath.github.io/agent-context-kit/skills/ - 2026-09-06 + 2026-09-22 https://cynrath.github.io/agent-context-kit/status/ - 2026-09-06 + 2026-09-22 https://cynrath.github.io/agent-context-kit/sync/ - 2026-09-06 + 2026-09-22 https://cynrath.github.io/agent-context-kit/trust-flow/ - 2026-09-06 + 2026-09-22 https://cynrath.github.io/agent-context-kit/verification/ - 2026-09-06 + 2026-09-22 https://cynrath.github.io/agent-context-kit/vscode/ - 2026-09-06 + 2026-09-22 https://cynrath.github.io/agent-context-kit/workflows/ - 2026-09-06 + 2026-09-22