spec

A short feature document that records what should be built, how success will be checked, and what is out of scope. It acts as a shared reference for implementation and review.

In plain words
What is it for?
Use it before substantial feature work to write an intent, acceptance checklist, exclusions, constraints, verification plan, and open questions in specs/.
Why use it?
It prevents unclear requests and scope creep by turning the intended result into specific, testable conditions before coding begins.

Command

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add commands/dshakes/compass/spec
Clone the repo
git clone --depth 1 https://github.com/dshakes/compass
Per session 44 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 661 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00044 $0.00661
Opus 5 $0.00022 $0.00331
Sonnet 5 $0.00009 $0.00132
Haiku 4.5 $0.00004 $0.00066

Measured 2d ago against content hash a603ce512696, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

spec scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

claude/commands/spec.md · 40 lines

How it starts

The opening of the file, as written. The whole thing — 40 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Draft a spec / intent for: ${ARGUMENTS:-the task we're about to do}.

The point: give the loop a ground truth. Tests prove the code works; the spec proves it does what was asked. The Builder implements against it; the Reviewer and QA verify against its acceptance criteria. Keep it short and honest — a spec that rots is worse than none.

Do

  1. Read the relevant code, CLAUDE.md, and any linked issue first (don't spec in a vacuum).
  2. Write specs/<kebab-slug>.md with exactly these sections, tight:
    • Intent — 1–3 sentences: what outcome, for whom, why now.
    • Acceptance criteria — a numbered, verifiable checklist (each item testable; prefer "given/when/then" where it helps). This is the contract.
    • Non-goals — what this explicitly does NOT do (prevents scope creep).
    • Constraints / invariants — perf, security/tenancy, back-compat, dependencies.
    • Verification plan — how each acceptance criterion will be checked (test, manual, metric).
    • Open questions — anything needing a human decision before/while building.
  3. If a load-bearing invariant changes, say "needs ADR" and stop short (use /adr).
  4. Keep it under ~1 page. Commit it (git add specs/… && git commit). Report the path.

Then (spec-driven loop)

Hand the spec to implementation: ~/compass/sdlc/orchestrate.sh "<task>" with SDLC_SPEC=specs/<slug>.md set — the Planner plans to the spec, the Builder implements it, and the Reviewer is told to verify the diff against the acceptance criteria (flag any unmet or out-of-scope). The human approves the spec (intent), not just the diff.

End to end (cloud loop too): commit the spec in the same PR (it lives under specs/), or add a Spec: specs/<slug>.md line to the PR description. The GitHub Reviewer (sdlc-review.yml) detects the spec, verifies the diff against its Acceptance Criteria, and marks unmet criteria or out-of-scope changes as Blocking — so the auto-fix loop converges on intent, not just "tests pass." Same human merge gate.

Read the full file on GitHub · 40 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 40 lines · 44 tokens per session scan A a603ce512696

Subscribe to this mod's changes

spec is a command published in the GitHub repository dshakes/compass (19 stars, last pushed 8d ago), licensed MIT. It adds 44 tokens to every session and 661 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.