Menu
SkillsMP has collected 237 skills from adnan-drina/rhoai3-coding-demo. Open a skill to review its source and details.
Showing 40 of 237 collected skills.
Use at M1 ANALYZE as the kind index. Pin only --skill paved-road-m1. Follow steps.json in order (skill_view subskills; native kanban_attach.py). Happy-path terminator is kanban_request_review, not kanban_complete. kanban_block for external/platform (MaaS 500,…
Use at M2 PLAN as the kind index. Pin only --skill paved-road-m2. Follow steps.json in order (skill_view speckit-specify then plan/tasks, plan-migration-partition producer, check-spec-readiness, native check-partition-coverage.py and…
Before kanban_create — lint SDD specs, story bodies (typed BODY_* codes), and partition coverage (HTTP 1:1 via story.endpoints as METHOD /path, dest_file 1:N with supersede; refuse invented HTTP paths vs inventory; K4 dest_file round-trip REFUSE dest-9…
Use at dest-init to mint M1 ANALYZE and M2 PLAN together with --idempotency-key so a repeated postStart is a no-op. Do not mint M3 or M4. Do not kanban daemon --force. The RHDH autoStartMigration checkbox defaults true now that this consumer exists; off skips…
Use when authoring or reviewing SOUL.md identity files for Hermes seats in stage 080, choosing between SOUL.md, AGENTS.md, and /personality overlays, or wiring custom personalities. Do NOT use for project context files and AGENTS.md discovery mechanics…
Use when designing or reviewing Hermes Agent configuration for stage 080: config.yaml locations and precedence, main and auxiliary model wiring, context_length and max_tokens, compression settings, provider options, fallback chains, and profiles. Do NOT use…
Use when M1 must inventory legacy@3.x entry points and the types reachable from those files. Writes evidence/entry-point-inventory.json then evidence/type-inventory.json from harvest_referent in evidence/derived/legacy-at-3.json (do not hardcode…
Runs kantra against the legacy@3.x referent and emits normalized findings plus the bounded M2 handoff. Use at M1 ANALYZE, at M5 for the findings delta, or when findings-handoff.json is missing, stale by digest, or failing schema validation.
Use at M2 PLAN — before minting Kanban children — to follow the Hermes skills speckit-specify, speckit-plan, and speckit-tasks (never implement), read M1 attachments, author evidence/partition.json, convert with k4_convert.py --partition --tasks, and mint…
Use before writing M4 PROVISIONAL_ACCEPT — refuse unless every gate skill pinned on the M4 card has a runner receipt under evidence/receipts/gates/ with argv, rc, and producer. ran true under evidence/verdicts/, a missing receipt, or M4 write_file of the…
When a workspace has no .specify/ — installs pinned Spec Kit, the Non-Goals override, the unique-owner tasks-template override, the destination constitution, the speckit overlay that removes implement, the AD-S stop rule, copies speckit-specify into the…
Use whenever creating a new Agent Skill or modifying an existing one in this project — writing or editing any SKILL.md, choosing what belongs in the body vs references/, structuring scripts and assets, or deciding whether a capability should be a skill at…
Use when writing or revising the description field of any SKILL.md, or when a skill fires on the wrong prompts or fails to fire on the right ones. Covers the official agentskills.io guidance: imperative pushy phrasing, the 1024-character limit, trigger eval…
Use when proving a skill actually improves agent output — before landing a substantial new skill, when iterating on one that underperforms, or when deciding whether a skill earns its context cost. Covers the official agentskills.io eval loop: test cases,…
Use when explaining Hermes Agent, choosing which Hermes mechanism fits a stage 080 need (skills vs memory, kanban vs delegate_task vs /goal, hooks vs plugins vs cron), checking platform/install support tiers, or verifying a claimed capability exists. Do NOT…
Use when invoking or scripting the Hermes CLI in stage 080: the complete hermes subcommand tree, global flags, headless/CI usage (-z, chat -q, --json, exit codes), the slash-command registry across CLI/TUI/messaging surfaces, and shell completion. Do NOT use…
Use when designing or reviewing Hermes subagent delegation for stage 080: delegate_task mechanics (context isolation, tool inheritance, parallelism, depth limits, stall detection, steering), the Subagent-Driven Development (SubDD) methodology, and when to…
Use when designing or reviewing Hermes Persistent Goals (/goal) for stage 080: the judge loop, completion contracts, quality gates, background-process parking, turn budgets, and when to choose /goal vs a kanban card vs kanban --goal. Do NOT use for board…
Use when designing or reviewing Hermes event hooks for stage 080: choosing among the four hook systems (gateway, plugin, shell, outbound webhooks), the 24-event plugin-hook catalog, firing semantics and ordering, blocking and context injection, consent for…
Use when designing or reviewing Hermes Kanban orchestration for stage 080: task lifecycle and states, dispatcher and worker-lane behavior, retry and recovery paths, orchestrator fan-out, collaboration patterns, boards and attachments, dashboard/notifications,…
Use when working with Hermes Managed Scope in stage 080: admin-tier config and secret pins under /etc/hermes, pin precedence over user seats and shell, fleet secret handling via vault sources (Bitwarden, 1Password, command helper), egress credential…
Use when presenting a decision, proposal, or escalation in stage 080 work — kanban block escalations, review findings, ADR proposals, design choices with trade-offs: structure it as One-Three-One (1 problem, 3 options, 1 recommendation) per the official…
Use when designing or governing Hermes' self-improvement loop for stage 080: dest-init honest-off vs the official governed-on fleet recipe, the post-turn background review that writes memory and skills, the product off-switch…
Use when working with Hermes sessions in stage 080: session identity and storage (state.db), resume/handoff/search, retention and pruning, checkpoints and rollback, and building provenance/audit trails on session ids. Do NOT use for compression config (use…
Use when authoring or governing skills for Hermes worker seats in stage 080: SKILL.md format and frontmatter, discovery paths and precedence (bundled, user, external_dirs, plugin), skill bundles, write gates, curator lifecycle, Skills Hub trust levels, and…
Use when configuring or reviewing Hermes tools and toolsets for stage 080: the toolset model (bundles, composites, platform presets, dynamic MCP/plugin toolsets), per-profile/per-session tool restriction, terminal backends, background process management, and…
Manage shared agent guidance for this project — AGENTS.md, shared rules, shared skills, hooks, and tool bridges. Use when the user asks to create a rule, update a skill, audit rules, review skills, add a hook, create a subagent, or asks about .agents/ or…
Use when using the MTA CLI for application analysis, migration assessment, and modernization. Do NOT use for UI workflows (use mta-ui), VS Code (use mta-vscode), IntelliJ (use mta-intellij), or AI features (use mta-lightspeed).
Use when installing MTA 8.2 operator, UI, and CLI on OpenShift, including prerequisites, namespace setup, and verification. Do NOT use for using tools (use mta-cli/mta-ui) or AI features (use mta-lightspeed).
Use when using the MTA IntelliJ IDEA plugin for analyzing applications and resolving migration issues, including installation, run configuration, issue review, and Quick Fix resolution. Do NOT use for CLI analysis (use mta-cli), web UI workflows (use mta-ui),…
Use when configuring or using Red Hat Developer Lightspeed for MTA to modernize applications with AI assistance, including LLM provider configuration, code migration suggestions, and AI-assisted analysis. Do NOT use for installing MTA (use mta-install) or…
Use when checking MTA 8.2 release notes, new features, bug fixes, known issues, and deprecated features. Do NOT use for installing (use mta-install), using tools (use mta-cli/mta-ui), or AI features (use mta-lightspeed).
Use when creating custom rules for MTA 8.2 analysis, including YAML rule syntax, rule metadata (labels, categories, effort), provider conditions (java, builtin, go, csharp, python, nodejs), logical conditions, condition chaining, custom variables, rule…
Use when using the MTA web UI for application assessment, analysis, migration waves, and issue management. Do NOT use for CLI (use mta-cli), VS Code (use mta-vscode), IntelliJ (use mta-intellij), or AI features (use mta-lightspeed).
Use when using the MTA VS Code extension for identifying and resolving migration issues in the IDE. Do NOT use for CLI (use mta-cli), UI (use mta-ui), IntelliJ (use mta-intellij), or AI features (use mta-lightspeed).
Refuse a worker log in which the agent routed around a tool refusal instead of reporting it (AD-020 §detector).
Use on the M3 stamp card (--check-only) and again before writing M4 PROVISIONAL_ACCEPT — refuse unless src/ and pom.xml are committed against HEAD. An M3 story-complete is not the harvest; the stamp card is. M4 still refuses a dirty tree. Do not dest-push. Do…
Measures the destination against the referent and emits G-1..G-4 ACCEPT/REFUSE/INCONCLUSIVE verdicts. Use at M4 for volume and substance, at M5 for findings delta and runtime parity, or when pinning a kill ratio or re-proving an edited evaluator.
Use before advancing or shipping — lint M4/M5 verdict tokens and floor receipts, parse surefire XML, snapshot test reports before any rebuild, and refuse an M4 card body that already names a verdict token or ship flag. Do not compose…
Use at M4 VERIFY to compose evidence/verdicts/m4-verdict.json from measured floor exit codes, including an explicit failed_floors field. Use when writing the M4 verdict or creating the M4 card, even if the user does not name a schema. Do not record a failed…