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 skills add tzachbon/smart-ralph --skill ralph-specum-cancelgit clone --depth 1 https://github.com/tzachbon/smart-ralphWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/tzachbon/smart-ralph/ralph-specum-cancel)<a href="https://agentmods.dev/skills/tzachbon/smart-ralph/ralph-specum-cancel"><img src="https://agentmods.dev/badge/skills/tzachbon/smart-ralph/ralph-specum-cancel/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/tzachbon/smart-ralph/ralph-specum-cancel"><img src="https://agentmods.dev/badge/skills/tzachbon/smart-ralph/ralph-specum-cancel.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00045 | $0.00930 |
| Opus 5 | $0.00023 | $0.00465 |
| Sonnet 5 | $0.00009 | $0.00186 |
| Haiku 4.5 | $0.00005 | $0.00093 |
Grade A, and why
ralph-specum-cancel scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 12d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 40 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Ralph Specum Cancel
Use this to stop execution safely and optionally remove a spec. Safe cancel preserves the spec and every prototype source path.
Derive RALPH_CODEX_PLUGIN_ROOT from this loaded skill by resolving two parent directories from the SKILL.md directory. Never derive it from the project working directory.
Contract
- Resolve the target by explicit path, exact name, or
.current-spec - Treat a missing
activePrototypesfield as an empty map - Publish reviewed immutable cancellation evidence before removing active state
- Confirm the exact local target before any deletion
- Never delete a remote branch or perform another remote action
- Do not guess on ambiguous names
Action
- Resolve the target spec. If none exists, report that there is nothing to cancel.
- Read
.ralph-state.jsonwhen present and summarize the current phase and progress. - If
activePrototypesis non-empty, use the resolvedbasePathand run"$RALPH_CODEX_PLUGIN_ROOT/scripts/prototype_records.py" reconcilebefore cancellation. Process active IDs bycreated, then ID, at the next safe tool boundary. - For each active prototype:
- Inspect its
owner,leaseToken, andharnessRun.id. WhenownerandleaseTokenare both null or absent, verify thatharnessRun.idis also null or absent and no builder is associated, then skip interrupt and release and continue cancellation. If those fields disagree, treat the entry as inconsistent builder ownership: retain the lease and active entry and stop for that ID. Otherwise, interrupt only its recorded child builder through the bounded harness contract. Release throughlocked_state.py release-lease --id <id> --lease-token <leaseToken>only after the harness verifies that the recorded builder and descendants stopped. If interruption is unavailable or unverified, or reports any failed termination attempt, retain the lease and active entry and stop cancellation for that ID. - Preserve its question, blocker, return phase and task, timestamps, local branch, isolation pointers, partial implementation, run evidence, stale metadata, and downstream artifacts.
- Render an exclusive terminal candidate with
verdict: cancelled,gateApproved: false, andsourceDisposition: retained. Never overwrite existing candidate or final bytes; allocate a new publication ID withsupersedes: ["<original-id>"]on collision. - Before review, store the exact returned
candidateHash. Update the existing original entry through compare-and-settransition, or reserve a separate superseding ID through create-onlylocked_state.py upsert-prototypewith the copied recovery fields and exactcandidateHash. Retain the original active entry unchanged while the superseding record is reviewed. A reservation collision preserves both entries and candidates and allocates another ID. - Send the exact candidate bytes and source pointers to the reviewer. Continue only after
REVIEW_PASS, then callreview-candidateandpublishwith the publication ID and exact candidate hash. - Re-read and parse the immutable final and verify its hash. Publication removes its own active publication entry. When a superseding ID was used, remove the original active entry only after final verification succeeds; any earlier failure retains it. Reconcile, then restore the recorded main phase and task index through
locked_state.pywithout changing unrelated state.
- Inspect its
- Safe cancel is the default. After every active prototype has a verified immutable
cancelledrecord, delete only.ralph-state.jsonthroughlocked_state.py delete-state. Keep the spec directory,.current-spec, terminal records, prototype source, partial work, scratch paths, worktrees, and local branches. - If full removal was explicitly requested, show the resolved spec directory and every prototype isolation path and branch. Require confirmation naming the exact spec directory before removing it. Prototype paths and local branches require separate exact confirmations and are never included in recursive spec cleanup. Never delete a remote branch.
- If no state exists, report that no active loop exists and remove nothing unless the exact full-removal confirmation is complete.
- Keep active epic files unless the user separately requests their removal. Report exactly what was removed and what was preserved.
What ships with it
1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 12d ago First seen · 40 lines · 45 tokens per session scan A e43cbacc5825
ralph-specum-cancel is a skill published in the GitHub repository tzachbon/smart-ralph (536 stars, last pushed 2d ago), licensed MIT. It adds 45 tokens to every session and 930 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.
Other skills, from other repositories
remember
Review reusable project knowledge and decide what belongs in project memory, notepad, or durable docs.
docx-comment-reply
Reply to comments (批注) in Word .docx/.doc files: extract comment context, draft replies, write threaded replies back, and validate OOXML.
explaining-machine-learning-models
Explain trained machine learning models through feature attribution, local explanations, and behavior summaries. Use as an explicit/manual helper once a model already exists, not for training ownership, leakage auditing, or general ML strategy selection.
moai-ref-ui-polish
UI polish and interface-completion reference: the small visual details — concentric border radius, optical alignment, shadow-vs-border, motion easing, typography smoothing, tabular numbers, icon stroke weight, hit areas — that separate polished interfaces from generic ones. Agent-extending skill that amplifies…
config-validator
Validate AIWG configuration files and project setup for correctness and completeness.
aiwg-regenerate-copilot
Regenerate copilot-instructions.md for GitHub Copilot with vendor-specific content only.