PreToolUse
25Hook
Runs before the agent uses a tool for ^(Bash, PowerShell, Edit and Write)$ tool calls, running node. From V-Songbird/razor.
Hook
Runs before the agent uses a tool for ^(Bash, PowerShell, Edit and Write)$ tool calls, running node. From V-Songbird/razor.
Hook
Runs when the agent finishes a response, running node. From V-Songbird/razor.
Hook
Runs when you submit a prompt, before the agent sees it, running node. From V-Songbird/razor.
Skill Claude CodeCodex
Audits a project's manifest for dependencies no source file imports — the reverse of razor's write-time gates (which block NEW dependencies), this finds EXISTING dead weight. Report-only; never edits any file.
Plugin Claude Code
Claude Code plugins by Victor Villegas — tools that make Claude plan better, talk less, build leaner, and stop guessing.
Hook Claude Code
Runs after a tool call finishes for ^(Edit, Write)$, ^(Bash and PowerShell)$ tool calls, executing run-tests-on-edit.js, nudge-manifest-curator.js and check-checkpoint-commits.js via node (3 commands). From V-Songbird/foundry.
Settings file Claude Code
Agent settings declaring 1 hook event (PostToolUse).
Plugin Claude Code
Experimental agentic plugins and tools that are not meant for public consumption.
Hook Claude Code
Runs after a tool call finishes for ^(Edit and Write)$ tool calls, executing run-tests-on-edit.js and nudge-manifest-curator.js via node (2 commands). From V-Songbird/slag.
Settings file Claude Code
Agent settings declaring 1 hook event (PostToolUse).
Instructions file CodexOpenCode
Instructions for V-Songbird/slag, a project described as: A sandbox marketplace of plugin experiments.
Plugin Claude Code
Analyzes the instruction files an agent loads — CLAUDE.md, .claude/rules/, AGENTS.md, skills and hooks — flagging what is vague, stale, buried, or better served by a mechanism, and rewriting the weak ones. /assay:opus5, /assay:sonnet5, /assay:haiku45 and /assay:fable5 audit this host, one per model; /assay:codex…
Skill Claude CodeCodex
Audits the instruction system Codex loads for a project — the AGENTS.md chain it reads, the .agents/skills it lists, the hooks around them — and reports what never takes effect, with reviewable fixes. A file shadowed by another in its own directory, or past the host's size limit. A path that no longer exists. Two…
Skill Claude CodeCodex
Audits the instruction system Codex loads for a project — the AGENTS.md chain it reads, the .agents/skills it lists, the hooks around them — and reports what never takes effect. A file shadowed or past the size limit. A path that no longer exists. Two rules that contradict each other. Findings only, with no letter…
Skill Claude CodeCodex
Interviews you about the behavior you want enforced, then writes one rule into the instruction file Claude Code actually loads — CLAUDE.md or a scoped .claude/rules/ .md — after checking the draft against the rules already active, and applies it as a reviewed, reversible change. Use when the user wants a rule written…
Skill Claude CodeCodex
Builds one Claude Code skill — .claude/skills/ /SKILL.md, with the format fetched live from the official docs — and installs it as a reviewed, reversible change. Also refits an existing skill's description that never seems to fire. Use when the user wants a skill created or made to trigger reliably — e.g. "make me a…
Skill Claude CodeCodex
Grades every rule in the project's CLAUDE.md and .claude/rules/, and every project skill description in .claude/skills/, exactly as /assay:opus5 does — the same files, the same report, the same offer to rewrite what is weak. The one difference is that the findings are written for Claude Fable 5. Use when the user…
Skill Claude CodeCodex
Grades every rule in the project's CLAUDE.md and .claude/rules/, and every project skill description in .claude/skills/, exactly as /assay:opus5 does — the same files, the same report, the same offer to rewrite what is weak. The one difference is that the findings are written for Claude Haiku 4.5. Use when the user…
Skill Claude CodeCodex
Grades every rule in the project's CLAUDE.md and .claude/rules/ on how it is written, scoped and placed, and flags the ones a hook or a skill would enforce better than prose. Also grades each project skill's frontmatter description in .claude/skills/ against the trigger recipe, and offers to rewrite what it finds.…
Skill Claude CodeCodex
Undoes one completed assay run in full, putting every file it wrote back the way it was — restoring the ones it edited and deleting the ones it created — from the run's recorded commit or from its copies under .assay/backup- /. Reads the run record in .assay/transactions.jsonl and refuses the whole undo rather than…
Skill Claude CodeCodex
Grades every rule in the project's CLAUDE.md and .claude/rules/, and every project skill description in .claude/skills/, exactly as /assay:opus5 does — the same files, the same report, the same offer to rewrite what is weak. The one difference is that the findings are written for Claude Sonnet 5. Use when the user…
Plugin Claude Code
Watches how full the context window is getting and, once you near the edge, nudges you to run /compact with a ready-made instruction — so the summary keeps the current task, decisions, and errors instead of an automatic guess.
Hook
Runs when you submit a prompt, before the agent sees it, executing brink.js via node. From V-Songbird/slag.
Plugin Claude Code
Routes Claude's reads, searches, and edits through your running JetBrains IDE for live error detection, unsaved-change visibility, and index-aware search.