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/koldovsky/project-factory/initgit clone --depth 1 https://github.com/koldovsky/project-factoryWhat 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 | $0.00045 | $0.02680 |
| Opus 5 | $0.00023 | $0.01340 |
| Sonnet 5 | $0.00009 | $0.00536 |
| Haiku 4.5 | $0.00005 | $0.00268 |
Grade A, and why
init scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 165 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/project-factory:init — install the loop
Install the framework's per-project loop into the current repository by
copying bundled files from the plugin (${CLAUDE_PLUGIN_ROOT}) and wiring them
up. This is Gate G0: no feature work before the loop that guards it exists.
The app stack itself (e.g. create-next-app) is NOT installed here — that is a
later orchestrator step after the stack decision; init is stack-agnostic.
Flags in $ARGUMENTS: --tools= selects which tool adapters to install
(default all — see step 10); --with-automations also installs the cloud
automations workflow; --force overwrites framework-owned files instead of skipping.
Hard rules
- Idempotent + non-destructive. For every target: if it exists, SKIP and
report it (unless
--force); never clobber the user's code or config. - Merge, don't overwrite, structured config: for
package.jsonand.claude/settings.json, ADD missing keys only; keep everything already there. - Report at the end: a table of what was added vs skipped.
Steps
-
Repo + dirs.
git initif not a repo. Create (if absent).claude/,scripts/,scripts/automations/lib/,docs/,docs/adr/,docs/qa/waivers/,evals/cases/,evals/results/,.githooks/,.github/workflows/,openspec/,quality/,trace/,retro/corrections/,templates/retro/..gitignoremust NOT ignoredocs/qa/**— QA evidence (videos, screenshots, reports) is committed PROOF. Ignore only temp:node_modules/,coverage/, and recording**/raw/scratch dirs. If a broaderdocs/qaignore exists, fix it. -
Agents & workflows (copy verbatim, skip-if-exists):
${CLAUDE_PLUGIN_ROOT}/agents/*.md→.claude/agents/(the plugin already provides these natively, but deliver project-local copies so the workflows resolve them by bareagentTypename)${CLAUDE_PLUGIN_ROOT}/.claude/workflows/*.js→.claude/workflows/
-
Deterministic scripts (copy + drop the
.referencesegment):scripts/check-traceability.reference.mjs→scripts/check-traceability.mjs(same forcheck-coverage-ratchet,check-eval-ratchet,check-trajectory,check-a11y,gate-status,qa-verify).scripts/record-demos.reference.mjs→scripts/record-demos.mjs(the headless recording+validation harness — demo and bugfix proof viaOUT_DIR);scripts/check-recordings.reference.mjs→scripts/check-recordings.mjs.- Reflection layer (same copy-and-rename rule):
check-acceptance-methods,check-visual-fidelity,check-process-ratchet,check-factory-integrity,ledger,ledger-report,correct— eachscripts/<name>.reference.mjs→scripts/<name>.mjs. These implement the honesty checks (acceptance join, vacuity semantics, stub scan, integrity lock) — they are part of the loop, not optional extras. scripts/automations/{run,drift-watch,dep-audit,ci-triage}.reference.mjs→scripts/automations/*.mjs;scripts/automations/lib/*.mjsverbatim;automations/registry.json→automations/registry.json(ships OFF).
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 · 165 lines · 45 tokens per session scan A 9f4dad02ff1f
init is a command published in the GitHub repository koldovsky/project-factory (4 stars, last pushed 1mo ago), licensed MIT. It adds 45 tokens to every session and 2,680 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 2026-08-31.
Other commands, from other repositories
checklist
Generate a custom checklist for the current feature based on user requirements.
clarify
Identify underspecified areas in the current feature spec by asking up to 5 highly targeted clarification questions and encoding answers back into the spec.
specify
Create or update the feature specification from a natural language feature description.
analyze
Perform a non-destructive cross-artifact consistency and quality analysis across spec.md, plan.md, and tasks.md after task generation.
converge
Assess the current codebase against the feature's spec, plan, and tasks, then append any remaining unbuilt work as new tasks to tasks.md so implement can complete it.
implement
Execute the implementation plan by processing and executing all tasks defined in tasks.md.