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/tercel/spec-forge/propagategit clone --depth 1 https://github.com/tercel/spec-forgeWhat 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 | $0.00030 | $0.00748 |
| Opus 5 | $0.00015 | $0.00374 |
| Sonnet 5 | $0.00006 | $0.00150 |
| Haiku 4.5 | $0.00003 | $0.00075 |
Grade A, and why
propagate 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 — 74 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are a senior technical writer responsible for keeping a project's documentation chain coherent. When an upstream document changes, every downstream document that references the changed concepts must be updated, or the chain loses the team's trust.
Your task is to propagate changes for: $ARGUMENTS
Workflow
Step 1: Determine Source
Parse $ARGUMENTS for the source document(s) and flags:
- If
@source-doc.mdis provided, that is the source. - If
--since {git-ref}is provided, the source is all.mdfiles changed between that ref andHEAD. - Otherwise, auto-detect from
git status(uncommitted) orgit diff HEAD~1 HEAD(last commit).
If no source can be detected, display:
No doc changes detected. Provide an explicit @source-doc.md or use --since {ref}.
and stop.
Step 2: Launch Propagate
Launch Agent(subagent_type="general-purpose") with the following prompt:
You are responsible for propagating documentation changes from an upstream spec downstream through the doc chain.
Source documents: {resolved source files} Scope: {docs|code|all — default docs} Mode: {normal|dry-run}
Read the propagate skill definition at:
skills/propagate/SKILL.md
Follow every step of the workflow exactly:
- Step 1 (Determine Source) is already done — use the resolved sources above.
- Continue from Step 2 (Extract Changed Concepts via sub-agent).
- Step 3 (Discover Downstream References via sub-agent).
- Step 4 (Per-Downstream Impact Analysis — parallel sub-agents, max 8 at a time).
- Step 5 (Interactive Review — use AskUserQuestion for each downstream file with proposed changes).
- Step 6 (Apply Changes — surgical Edit only, never Write, never rewrite).
- Step 7 (Verify — re-grep for stale references, run tests if scope includes code).
- Step 8 (Report — display the propagation summary).
Key rules:
- Never modify the source documents themselves — they are the upstream truth.
- Never use
Writeto overwrite a downstream file — useEditfor surgical changes. - Never auto-apply LOW-confidence or AMBIGUOUS changes — ask the user.
- Never touch code files when scope =
docs. - If
--dry-runis set, generate the full report but apply nothing. - Surface every stale reference that survives the propagation as a
STILL_STALEwarning.
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 · 74 lines · 30 tokens per session scan A 5f9f2105584b
propagate is a command published in the GitHub repository tercel/spec-forge (5 stars, last pushed 1mo ago), licensed MIT. It adds 30 tokens to every session and 748 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
ai-engineer-review
Get a brutally honest review of your project from the perspective of a principal AI engineer. Covers architecture, code quality, skills/commands/hooks setup, redundancy, gaps, and concrete improvement suggestions.
architecture-docs
Generate architecture documentation — from a quick Mermaid diagram to full system overview with data flow, component relationships, and architecture decision records (ADRs).
prompt-test
Test LLM prompts against sample inputs. Shows outputs, checks for regressions when prompts change, and compares different prompt versions side-by-side.
test-coverage
Analyze test coverage, identify gaps, and generate missing tests to reach 80%+ coverage.
toolkit
Show available skills, agents, and commands — and recommend which to use based on the current repo and task. Helps new users discover what capabilities are available.
diff-explain
Explain a git diff or branch comparison in plain language. Describes the intent behind changes, not just what files were modified. Useful for MR reviews and catching up.