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.
npx agentmods add commands/clownware/product-dev/compilegit clone --depth 1 https://github.com/clownware/product-devWhat 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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00010 | $0.01199 |
| Opus 5 | $0.00005 | $0.00600 |
| Sonnet 5 | $0.00002 | $0.00240 |
| Haiku 4.5 | $0.00001 | $0.00120 |
Grade A, and why
compile scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 95 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Compile the current project's artifacts into a spec package.
Gate Check
-
Read
.product-dev/context.json(or the directory specified in$ARGUMENTS). Verify it exists and has artifacts registered. -
Check minimum required artifacts:
- Context layer (at least one of):
problem_statement,solution_concept - Spec layer (at least one of):
data_models,user_flow
- Context layer (at least one of):
-
If minimum artifacts are missing:
"Not enough artifacts to compile. You need at least a problem statement and solution concept (run
/idea), plus data models and user flow (run/spec). Currently have: [list]." -
Report optional missing artifacts as warnings, not blockers:
proto_persona— persona will be absent from context layerhypothesis_statement— hypothesis will be absentscreen_inventory— screens.yaml will be skippedapi_contracts— endpoints.yaml will be skippedbusiness_rules— rules.yaml will be skippednfr— constraints.yaml will be skipped
Execution
-
Preflight — Python is required and not guaranteed to be present. The compile pipeline is a Python script. Before running it, check the environment:
python3 --version && python3 -c "import yaml" 2>/dev/null && echo PYYAML_OK- If
python3is missing, do NOT proceed. Tell the user:"Compiling the spec package requires Python 3, which isn't available in this environment. Your working artifacts in
.product-dev/artifacts/are complete and usable as-is, and/summaryproduces a consolidated prose brief without Python. Install Python 3 to enable the full validated spec package." - If
python3exists butPYYAML_OKdid not print, install the dependency:
If that install fails (nopip install -r "${CLAUDE_PLUGIN_ROOT}/scripts/requirements.txt" -qpip, no network), fall back to the same graceful message above — do not leave the user with a raw traceback.
- If
-
Run the compilation script:
python3 "${CLAUDE_PLUGIN_ROOT}/scripts/compile_spec.py" <project-dir>The script handles:
- Copying prose artifacts to
spec-package/context/ - Extracting YAML from spec artifacts to
spec-package/spec/ - Compiling a PRD from context artifacts to
spec-package/docs/prd.md - Extracting ADRs from scope boundaries and exclusions to
spec-package/docs/adrs/ - Generating
manifest.yamlwith reading order, defaults, and validation summary - Running validation (20 cross-reference checks) and writing
validation-report.yaml - Generating a
CLAUDE.mdhandoff instruction for the implementation agent
- Copying prose artifacts to
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.
- yesterday First seen · 95 lines · 10 tokens per session scan A 0aff9011a93e
compile is a command published in the GitHub repository clownware/product-dev (3 stars, last pushed 6d ago), licensed Apache-2.0. It adds 10 tokens to every session and 1,199 once invoked, about $0.0001 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 2026-08-31.
Other commands, from other repositories
start
Front door. Reads the current project (framework, tokens, brief, spec, harness) and reports what ui-craft can do right now, then routes you to the right next step. Run this first if you're new or unsure where to begin. No code changes.
tokens
Audit or establish the project's 3-layer token spine.
remember
Record a learned design constraint from a correction into the project brief.
craft
One-shot build pipeline for a complete surface from an outcome recipe — inputs (or defaults) → composition → theme → build order → acceptance bar. Use when the user asks for a whole surface ("build me a dashboard", "I need a landing page") and expects a shippable result, not component-level help.
audit
Technical UI audit — a11y, performance, responsive. Produces a prioritized findings table.
brief
Write or update the project's durable design brief at .ui-craft/brief.md.