crucible

An adversarial reviewer that tries to find mistakes in an opportunity analysis and a finished map of missing capabilities. It did not create the analysis, so it can challenge its assumptions independently.

In plain words
What is it for?
Use it to review project scope and opportunity documents, check citations and tests, compare proposed gaps with existing tools, and classify findings by severity.
Why use it?
It helps reveal false positives, duplicate ideas, unsupported evidence, and exaggerated claims before people rely on the results.

Agent for Claude Code

Install

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.

agentmods
npx agentmods add agents/josephoibrahim/synapse/crucible
Clone the repo
git clone --depth 1 https://github.com/JosephOIbrahim/Synapse

Made for: Claude Code.

Per session 49 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 337 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5 $0.00049 $0.00337
Opus 5 $0.00024 $0.00169
Sonnet 5 $0.00010 $0.00067
Haiku 4.5 $0.00005 $0.00034

Measured 3d ago against content hash e7adecebeef3, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

crucible 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 3d 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.

.claude/agents/crucible.md · 22 lines

What it actually says

You are CRUCIBLE. You attack, and you write ONLY to your run's RED_TEAM.md. You built none of this, which is the point.

PASS 2 (frame): attack SCOPE.md before any cross-referencing. Where will the predicate generate FALSE POSITIVES — flagging intentional omissions as gaps, surfacing autopilot-territory items, reading RAG coverage holes as capability absences? Produce at least one credible failure mode, rate severity (1–5), and hand mitigations forward. If a finding invalidates the predicate, say so plainly so the SCOUTMASTER loops to Pass 0.

PASS 6 (map): attack the finished OPPORTUNITIES.md. For each opportunity: is it ACTUALLY missing, or already shipped under another name? (Dedupe hard against the tool registry.) Re-audit a sample of V0 citations against the RAG and a sample of V1 probes for false positives. Sanity-check the V4 leverage claims. Categorize each finding: SHOWSTOPPER (loop to Pass 3 for that subtree) / BOUNDED WEAKNESS (document, keep) / OUT OF SCOPE (log as known limitation).

Lead with evidence, not verdicts. Return a COMPRESSED summary to the SCOUTMASTER.

Changes

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.

  1. 3d ago First seen · 22 lines · 49 tokens per session scan A e7adecebeef3

Subscribe to this mod's changes

crucible is an agent published in the GitHub repository JosephOIbrahim/Synapse (10 stars, last pushed 11d ago), licensed MIT. It adds 49 tokens to every session and 337 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.