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/habib0x0/spec-driven-plugin/spec-execgit clone --depth 1 https://github.com/Habib0x0/spec-driven-pluginWhat 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 | $0.00014 | $0.00476 |
| Opus 5 | $0.00007 | $0.00238 |
| Sonnet 5 | $0.00003 | $0.00095 |
| Haiku 4.5 | $0.00001 | $0.00048 |
Grade A, and why
spec-exec 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.
What it actually says
/spec-exec Command
Run a single iteration of spec-driven implementation. Claude picks the highest-priority pending task, implements it, tests it, updates the spec, and commits.
Usage
Run the script directly from your project root:
${CLAUDE_PLUGIN_ROOT}/scripts/spec-exec.sh [--spec-name <name>]
Or via Bash tool if invoked within Claude Code:
bash ${CLAUDE_PLUGIN_ROOT}/scripts/spec-exec.sh --spec-name <name>
Arguments
--spec-name <name>- Which spec to execute against. Auto-detected if only one spec exists in.claude/specs/.
What It Does
- Reads your spec files (requirements.md, design.md, tasks.md)
- Builds a prompt with the full spec context
- Runs
claude --dangerously-skip-permissions -pwith that prompt - Claude implements one feature, tests it, updates the spec, and commits
Prerequisites
- A completed spec in
.claude/specs/<name>/with all three files - Run
/spec <name>first if you haven't created a spec yet
Quick Preflight
Before running a single iteration:
# 1. Verify git is configured
git remote -v | head -1
# 2. Check the build is clean
npm run build 2>&1 | tail -3
# 3. Confirm spec exists
[ -d ".claude/specs/<spec-name>" ] && echo "✓ Spec ready"
Tips
- Run one at a time: Each exec picks ONE task. Run multiple execs to implement multiple tasks.
- Check progress.md after each: Tells you what was completed and what's next
- Use /spec-loop for batches: If you want to run many iterations automatically, use
/spec-loopinstead - Integration sweep runs at end: When all tasks complete, /spec-loop triggers mandatory final validation. Single /spec-exec iterations don't do this.
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.
- 2d ago First seen · 65 lines · 14 tokens per session scan A 7da2346b112b
spec-exec is a command published in the GitHub repository Habib0x0/spec-driven-plugin (10 stars, last pushed 3mo ago), licensed MIT. It adds 14 tokens to every session and 476 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
ship
Deliver the increment. Then go live with a rollback you have run.
sruja-bugfix-triage
Systematic workflow for triaging and fixing bugs with architecture grounding.
sruja-verify-task
Run verification steps for a task profile. Use before marking a task as done.
AI: Decide
Build-vs-adopt gate — for each critical concern, decide Rent/Adopt/Extend/Fork/Build and record it as an ADR.
re-review
Round 2+ review — confirm revisions landed, catch regressions, do not re-litigate.
xray
Zero-config X-ray of this repo's real architecture — map, living docs, and evolution feed. No rules required.