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.
git clone --depth 1 https://github.com/giggsoinc/ravenWrote 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/commands/giggsoinc/raven/raven-harden)<a href="https://agentmods.dev/commands/giggsoinc/raven/raven-harden"><img src="https://agentmods.dev/badge/commands/giggsoinc/raven/raven-harden/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/commands/giggsoinc/raven/raven-harden"><img src="https://agentmods.dev/badge/commands/giggsoinc/raven/raven-harden.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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 | $0.00061 | $0.00926 |
| Opus 5 | $0.00030 | $0.00463 |
| Sonnet 5 | $0.00012 | $0.00185 |
| Haiku 4.5 | $0.00006 | $0.00093 |
Grade A, and why
raven-harden 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 10d 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 — 151 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/raven-harden
Turns session observations into permanent Raven intelligence. Run when prompted by Task-Observer or whenever you want to solidify what's been learned.
Steps
1. Read the Log
Read: docs/observations/security_log.md
Find: all entries with Status: open
Find: all entries with Status: promotion-candidate
Count: total open, total candidates
Report summary before doing anything:
Raven Harden — Review
══════════════════════════════════════
Open observations: [N]
Promotion candidates: [N platform(s)]
Last hardened: [date or never]
Open observations:
[date] [platform] — [one-line summary]
...
Promotion candidates:
[platform] — used [N] times, [N] log entries
...
══════════════════════════════════════
What would you like to do?
[A] Harden all open observations
[S] Select which to harden
[P] Promote candidates to curated skills
[N] Nothing — just reviewing
Wait for user input before proceeding.
2. Harden Observations → CLAUDE.md
For each approved observation:
Assess where it belongs:
- Rule that should never be broken → add to CLAUDE.md Non-Negotiable Rules
- Platform-specific Giggso pattern → add to docs/knowledge/internal_raven_ops.md
- General engineering pattern → add to docs/knowledge/general_security_patterns.md
CLAUDE.md hardening — append to relevant section:
## Hardened Rules (from observations)
| Date | Rule | Source |
|---|---|---|
| [date] | [rule in one line] | [platform / observation type] |
Mark entry in log:
Status: hardened → [date]
3. Promote Candidates → Curated Skills
For each platform with Status: promotion-candidate:
1. Read all log entries for that platform
2. Read cache: .raven/.cache/dynamic-skills/[platform-slug].md
3. Assemble SKILL.md using accumulated profile
Template for promoted skill:
---
name: [platform]-specialist
description: [generated from log observations]
allowed-tools: Agent, WebSearch, Read, Write
---
# [Platform] Specialist — [Expert Name]
**Expert persona:** [from cache]
**Promoted from:** dynamic-specialist after [N] sessions
**Last updated:** [date]
## Core Focus
[from cache + log]
## Key Rules (from observations)
[all confirmed patterns from log entries]
## Known Gotchas (confirmed in practice)
[gotchas from log entries]
## Agent Chain
Step 1 → check guard agents relevant to this platform
Step 2 → [platform] expertise below
Step 3 → search agent ON DEMAND only
[rest of specialist content from cache profile]
---
4. Write to: core/skills/[platform]-specialist/SKILL.md
5. Mark log entries: Status: promoted → [date]
6. Commit to giggsoinc/raven and push
7. raven-sync automatically distributes to all registered projects
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.
- 10d ago First seen · 151 lines · 61 tokens per session scan A 1c53232d2a96
raven-harden is a command published in the GitHub repository giggsoinc/raven (5 stars, last pushed 9d ago), licensed MIT. It adds 61 tokens to every session and 926 once invoked, about $0.0003 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.