agentcohort 0.1.0

package/dist/templates/commands/fix-blockers.md

@@ -0,0 +1,36 @@
+---
+description: Fix ONLY the listed blockers, then verify. No unrelated changes.
+argument-hint: <paste the blocker list from a review>
+---
+
+# /fix-blockers — Targeted Blocker Resolution
+
+Resolve **only** the specific blockers listed in `$ARGUMENTS` (typically the
+BLOCKER/HIGH findings from `/review-diff`, `/dev-flow`, or `/perf-hunt`).
+
+## Pre-flight
+
+Restate each blocker as a discrete, checkable item. If the list is vague or
+empty, **stop and ask** for the explicit blockers — do not infer scope.
+
+## Pipeline
+
+1. **feature-implementer** — fix each listed blocker with the minimal change.
+   One blocker at a time; do not touch anything not on the list.
+2. **test-verifier** — run tests/typecheck/lint; add/adjust focused tests for
+   the fixed blockers; report real output.
+
+## Rules
+
+- **Only the listed blockers.** No refactors, no renames, no reformatting, no
+  "while I'm here" fixes, no unrelated code.
+- No API/schema/auth/security/persistence semantic change unless that change
+  *was itself the blocker* and is explicitly approved.
+- New issues discovered while fixing → reported, not fixed (route to
+  `/bug-audit`).
+- Each blocker must end with evidence it is resolved.
+
+## Output
+
+Per-blocker: what changed (`path:line`), the verification command, and its
+real result. Then recommend re-running `/review-diff` to confirm clearance.

package/dist/templates/commands/perf-hunt.md

@@ -0,0 +1,40 @@
+---
+description: Performance pipeline — measure/evidence, bottleneck, safe optimization, verify, perf review.
+argument-hint: <what is slow + the workload/metric if known>
+---
+
+# /perf-hunt — Measure → Bottleneck → Safe Optimize → Verify → Review
+
+Orchestrate the performance workflow for `$ARGUMENTS`. **Evidence before
+changes. No blind optimization.**
+
+## Pipeline
+
+1. **performance-hunter** — define "slow" (metric, workload, target), gather
+   measurements, rank bottlenecks, separate measured from hypothesized.
+2. **solution-architect** — *only if* the likely fix affects caching, data
+   flow, or architecture. Decide the boundary-safe approach. Otherwise skip
+   and say why.
+3. **perf-optimizer** — apply the smallest reversible, evidence-backed change;
+   measure before/after under the same workload; preserve behavior.
+4. **test-verifier** — run tests/typecheck/lint; confirm behavior unchanged;
+   report real output.
+5. **perf-reviewer** — verify the gain is real and representative, behavior is
+   preserved, any cache has sound invalidation, and assess new regression
+   risk. Verdict required.
+
+## Rules
+
+- No optimization without a **measured** bottleneck behind it.
+- **No behavior/output change.** Same inputs → same results.
+- **No caching/memoization without an explicit, correct invalidation
+  strategy** and a stated staleness bound.
+- One optimization per change; keep it reversible.
+- If the gain is marginal or risk outweighs it, stop and report — do not ship
+  a risky micro-win.
+- If `perf-reviewer` BLOCKs, summarize and recommend `/fix-blockers`.
+
+## Output
+
+Stage summary including before/after numbers, behavior-preserved evidence, and
+the reviewer's APPROVE/BLOCK verdict.

package/dist/templates/commands/review-diff.md

@@ -0,0 +1,39 @@
+---
+description: Run the final reviewer on the current diff/PR. Read-only verdict.
+argument-hint: [base ref or PR — defaults to the current branch diff]
+---
+
+# /review-diff — Final Reviewer Only
+
+Run the **final-reviewer** subagent on the current change.
+
+## What to review
+
+- If `$ARGUMENTS` names a base ref or PR, diff against that.
+- Otherwise review the working diff against the repository's base branch
+  (e.g. `git diff` / `git diff <base>...HEAD`).
+
+## Mandate (read-only)
+
+The reviewer must judge, with `path:line` evidence and explicit severity
+(BLOCKER / HIGH / MEDIUM / NIT):
+
+- **Correctness** — incl. edge/error paths and concurrency.
+- **Regression risk** — what existing behavior could break.
+- **Scope creep** — anything changed the task did not authorize.
+- **Security** — input trust, authz, injection, secrets.
+- **Data consistency** — partial writes, transactions, invariants.
+- **Tests** — is the changed risky behavior actually covered.
+
+## Rules
+
+- **No edits.** This command produces a verdict only.
+- Unauthorized API/schema/auth/security/persistence semantic change is at
+  least HIGH, default BLOCKER.
+- Missing test for changed risky behavior is a BLOCKER.
+- No rubber-stamp, no nitpick-only review.
+
+## Output
+
+`APPROVE` or `BLOCK`, the findings list, and exactly what must change before
+it can land. If BLOCK, recommend `/fix-blockers` with the blocker list.

package/package.json

@@ -0,0 +1,56 @@
+{
+  "name": "agentcohort",
+  "version": "0.1.0",
+  "description": "Install a principal/staff-level Claude Code AI software engineering organization (agents + workflows + routing rules) into any project with one command.",
+  "keywords": [
+    "claude",
+    "claude-code",
+    "agents",
+    "subagents",
+    "workflow",
+    "ai",
+    "cli",
+    "developer-tools",
+    "code-review",
+    "bug-audit"
+  ],
+  "license": "MIT",
+  "author": "agentcohort",
+  "type": "commonjs",
+  "bin": {
+    "agentcohort": "dist/cli.js",
+    "agent-force": "dist/cli.js"
+  },
+  "main": "dist/index.js",
+  "types": "dist/index.d.ts",
+  "files": [
+    "dist",
+    "README.md"
+  ],
+  "engines": {
+    "node": ">=18"
+  },
+  "publishConfig": {
+    "access": "public"
+  },
+  "repository": {
+    "type": "git",
+    "url": "git+https://github.com/Thiendekaco/agentcohort.git"
+  },
+  "homepage": "https://github.com/Thiendekaco/agentcohort#readme",
+  "bugs": {
+    "url": "https://github.com/Thiendekaco/agentcohort/issues"
+  },
+  "scripts": {
+    "clean": "node -e \"require('fs').rmSync('dist',{recursive:true,force:true})\"",
+    "build": "npm run clean && tsc -p tsconfig.json && node scripts/copy-templates.mjs",
+    "test": "vitest run",
+    "test:watch": "vitest",
+    "prepublishOnly": "npm run build && npm test"
+  },
+  "devDependencies": {
+    "@types/node": "^20.14.0",
+    "typescript": "^5.5.0",
+    "vitest": "^2.0.0"
+  }
+}