By alibassam
Savagely self-critiquing research framework. Evaluates a local repo's product/idea claims against web evidence and emits a no-bullshit markdown report.
Stage 5. Reads the synthesizer's draft with one goal — find every remaining piece of bullshit. Produces a kill-list of hedges, unsupported adjectives, T3-resting conclusions, omitted counter-evidence, shined findings. Never rewrites; only critiques.
Stage 3 specialist. Takes every claim from claims.md and verifies it against independent web evidence. Tags each claim Supported / Partially Supported / Contradicted / Unverifiable with source tier and confidence.
Stage 2 agent. Reads the repo ingest summary and produces a structured, neutral, fully-sourced list of every claim the repo makes — problem, target user, proposed solution, quantitative claims, qualitative claims, differentiation claims, hedges.
Stage 3 specialist. Enumerates how the idea breaks in practice — known failure patterns in the space, adversarial use cases, things the pitch omits, reasons comparable efforts died.
Stage 3 specialist. Hunts for existing solutions to the problem the repo claims to solve. Identifies incumbents, cheaper/free alternatives, and historical attempts. Writes findings with source tiers and confidence tags.
The structured-claim-list format produced by the claim-extractor agent. Defines the exact schema every downstream specialist consumes. Read before extracting claims from a repo.
The 8-stage pipeline driver for the savage-research plugin. Takes a local repo path, runs ingest → claim extraction → parallel specialists → synthesis → adversary → revision → self-audit → emit, and produces the final markdown report. Invoked by the /research slash command.
The canonical source-tier system (T1-T5) and the 7 hard tone invariants every agent in the savage-research pipeline must follow. Read this before producing any output in a research run.
Merge rules for the synthesizer agent that combines five specialist scratchpads into a draft report without softening findings. Read before synthesizing or revising drafts.
Uses power tools
Uses Bash, Write, or Edit tools
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
A Claude Code plugin that runs a savagely self-critiquing, multi-agent research pipeline on a local repository and emits a no-bullshit markdown report.
Point it at a local repo. It reads the pitch, extracts every claim, dispatches five specialist agents in parallel (prior-art, claim-audit, failure-modes, traction, framing), synthesizes their findings, runs an adversary that tries to kill the synthesis, revises against the kill-list, self-audits against hard invariants, and emits a report to <repo>/.research/.
/research <local-path-to-repo>
Example:
/research ./test-fixtures/bullshit-repo
Add this repo as a plugin marketplace and install, directly from Claude Code:
# Inside Claude Code:
/plugin marketplace add AliBassam/savage-researcher
/plugin install savage-research@savage-researcher
That's it. No clone, no per-project setup. The /research command is available in every Claude Code session on your machine.
To uninstall:
/plugin uninstall savage-research@savage-researcher
To verify it's loaded:
/plugin
savage-research should appear in the plugins list, with /research available as a command.
After /research <path>, check <path>/.research/:
YYYY-MM-DD-HHMM-<repo>.md is the final report.runs/YYYY-MM-DD-HHMM/ contains the full working trail: ingest summary, claim list, all five scratchpads, every draft version, kill-list, and audit results.npx claudepluginhub alibassam/savage-researcher --plugin savage-researchYOU.md — a living user intelligence profile. 50 specialist frameworks analyze how you think, communicate, decide, and build. One global profile at ~/.claude/YOU.md that Claude reads on every message across every project.
Savagely self-critiquing resume / CV / portfolio review framework. Evaluates a candidate's claims against 2026 job-market reality and emits a no-bullshit markdown report.
Launch a team of AI agents for deep parallel codebase research — causal understanding, not just coverage
Multi-type research with parallel researchers and trust-aware synthesis
Research sprint orchestrator for Claude Code. Structured research with claims, evidence tiers, and compiled output.
Multi-agent deep research plugin with parallel web searches and synthesis
Agents used for research across multiple data sources. Other plugins expect this one to be enabled.
Adversarial multi-agent pipeline for Claude Code. GAN-style loops where generators produce artifacts, discriminators validate them, and feedback drives convergence.