Plugin Claude Code
Jintech OMG development toolkit — SDLC skills (/ticket, /implement, /verify, /prepr, /pr, /grill-me, /debug) + code-review-graph & product-graph MCP servers + autonomous verify trust core + enforcement hooks.
Instructions file
Claude Code instructions for ahsan0444/jintech-omg-dev, covering jintech omg dev — agent instruction layer, model ladder, session discipline, subagent discipline and mcp search policy (omg repos only).
Agent
Executes a single approved plan step in an OMG repo — grep-first edit, TDD when a test file exists, OMG layer compliance baked in. Use for every implementation step in /implement and every warning-fix in /prepr fix. Returns a STATUS schema, never prose.
Agent
Read-only OMG codebase locator. MCP-graph-first search that returns file paths, line ranges, and unique grep strings — never file contents. Use for every codebase query, trace, lint run, or psql lookup inside /ticket, /debug, /grill-me, /prepr, and /pr. Cannot read or edit files — the no-file-reads rule is enforced by…
Agent
Isolated executor for the /verify trust core — runs the tiered verification harness, judges by assertions (never impression), and on behavioral failure runs a bounded implementation-only self-fix loop. Returns a STATUS schema, never prose. Use only from the /verify skill.
Command
Bootstrap the Agent OS data home + verify harness on this machine (product graph, registry, Playwright, auth capture). Run once per machine, or after cloning, to set up the /verify workflow.
Command
Generate vector embeddings for the code-review-graph to enable semantic search. Run once after code-review-graph build. Safe to re-run — only re-embeds changed nodes.
Command
Extract reusable patterns from the current session and save as a skill file for future use.
Command
Load the most recent session file from /.claude/session-data/ and resume work with full context from where the last session ended.
Command
Show skill-router match statistics from /.claude/logs/intent-router-.jsonl and prune logs older than 30 days. Use to spot misrouting intents and dead patterns before tuning skill-routing-manifest.json.
Command
Save current session state to a dated file in /.claude/session-data/ so work can be resumed in a future session with full context.
Hook
Runs when you submit a prompt, before the agent sees it, executing run-hook.py via python3. From ahsan0444/jintech-omg-dev.
Hook
Runs before the agent uses a tool for Grep, Bash and Bash tool calls, executing run-hook.py via python3 (2 commands). From ahsan0444/jintech-omg-dev.
Hook
Runs after a tool call finishes for Edit, Write, MultiEdit, NotebookEdit, mcp__.*(code-review-graph and product-graph)__ tool calls, executing run-hook.py via python3 (2 commands). From ahsan0444/jintech-omg-dev.
Hook
Runs when a session starts on startup and resume, executing run-hook.py via python3. From ahsan0444/jintech-omg-dev.
Skill Claude CodeCodex
Root cause analysis for bugs and unexpected behaviour. Traces execution paths, ranks hypotheses, confirms the culprit, and routes to /implement or /ticket. Run before /ticket when the problem is unclear. Use whenever you know something is broken but not why.
Skill Claude CodeCodex
Interview the user relentlessly about a plan or design until reaching shared understanding, resolving each branch of the decision tree. Use when user wants to stress-test a plan, get grilled on their design, or mentions "grill me". Saves output to .planning/ for /ticket to auto-load.
Skill Claude CodeCodex
Executes an approved investigation plan — applies code changes via subagents, maintains a change log, handles failures, and verifies all edits. Run after /ticket produces an approved plan.
Skill Claude CodeCodex
Run a real pgMustard query-plan review against DB function(s) added/modified in the current branch's dbscripts, using the pgMustard save/score API when a token is available (falls back to local EXPLAIN emulation otherwise), producing a plain-English review with a 0-5 tip score.
Skill Claude CodeCodex
Creates a Bitbucket draft PR from the current branch using the Bitbucket REST API. Fetches Jira ticket content, generates a Perl critic report when Perl files changed, and derives a change summary from the implementation plan or git diff. If a PR already exists, rebases off the destination branch and pushes.
Skill Claude CodeCodex
Pre-PR review — audits all branch changes against OMG coding standards. Runs Perl::Critic for Perl files, checks OMG layer conventions, validates templates, JS, and DB scripts. Reports blockers vs warnings before a PR is raised.
Skill Claude CodeCodex
Investigates a Jira ticket via subagent orchestration and produces a Plan Mode proposal for review. Stops at plan approval — run /implement to execute.