Menu
O SkillsMP coletou 35 skills de kostadis/mytools. Abra uma skill para revisar a origem e os detalhes.
Mostrando 35 de 35 skills coletadas.
Propose→review→apply mechanical-residue scrub for session_doc narration prose. A deterministic regex pass surfaces candidate residue (raw numbers on DC/AC/HP/damage/healing/feet/rounds/initiative, out-of-fiction table-speak, real player names as speakers) —…
Idioma do texto original: inglês
Quote-level consistency check on scene_extractions_new/ — flags VTT transcription errors, garbled phrases, and clarity issues in verbatim quote blocks, then proposes corrections for user approval. Invoke as /session-summary-consistency [session-dir].
Idioma do texto original: inglês
Run consistency checks at every LLM-pipeline boundary (gm-assist → session-summary → scene extractions → narration) with a human-review gate between stages. Use when the user invokes /staged-consistency [session-dir] and wants the multi-stage check rather…
Idioma do texto original: inglês
Clean up Otter/Zoom VTT transcripts for a D&D campaign — applies the known-misspellings glossary and prompts the user about unrecognised proper nouns. Invoke as /vtt-spell-pass [vtt-path].
Idioma do texto original: inglês
Put speaker names on a session recording when the transcripts carry no usable attribution — the single-room case, where everyone shared one microphone so Zoom labelled all 866 cues with the host's name, and the editor's diarization produced six anonymous…
Idioma do texto original: inglês
Rebuild a session transcript that has correct timings OR correct speakers but not both — the split-brain case where `*.speakers.vtt` carries fabricated timestamps and Zoom's live text export flips speaker mid-sentence, shredding one person's sentence across…
Idioma do texto original: inglês
Quote-level consistency check on scene_extractions_new/ or scene_extractions/ for VTT transcription errors, garbled phrases, proper-noun misspellings, and clarity issues in verbatim quote blocks. Use when the user asks for /session-summary-consistency…
Idioma do texto original: inglês
Render corrected scene-extraction quote blocks into readable, in-voice prose in a derived scene_extractions_smoothed/ layer. Use when the user asks for /voice-smooth [session-dir] or wants voice-aware smoothing after session-summary-consistency.
Idioma do texto original: inglês
Clean up Otter/Zoom VTT transcripts for a D&D campaign by applying the known-misspellings glossary and asking the user to adjudicate unrecognised proper nouns. Use when the user asks for a VTT spell pass or invokes $vtt-spell-pass with a transcript path.
Idioma do texto original: inglês
Use when reviewing or editing D&D campaign narrative chapter markdown generated from session transcripts, especially when the user provides a campaign root path and a current summary/chapter file. Loads per-campaign canon docs, party state, planning pressure,…
Idioma do texto original: inglês
Run a CampaignGenerator consistency check on one session document, then review and apply only approved fixes. Use when the user asks to check a recap, enhanced summary, scene extraction, or narration for campaign consistency, including when they type…
Idioma do texto original: inglês
Run CampaignGenerator consistency checks at each session-doc pipeline boundary: gm-assist, session-summary, scene extractions, and narration. Use when the user asks for staged consistency or types /staged-consistency [session-dir].
Idioma do texto original: inglês
Render the verbatim scene-extraction quotes into readable, in-voice prose — a derived scene_extractions_smoothed/ layer that sits between /session-summary-consistency and session_doc. Uses each character's voice file as the guardrail. Fixes garble, run-ons,…
Idioma do texto original: inglês
Critique a session_doc.py narration for generic prose, voice drift, doc-level budget breaches, or conflicts with the character's voice spec and the campaign's genre rulebook. Use after generating per-scene narration or an assembled fable doc, before assembly…
Idioma do texto original: inglês
Review and merge near-synonym file-to-file link relations produced by drive-tagger. Interactive, one relation family at a time. Invoke as /drive-consolidate-links.
Idioma do texto original: inglês
Load the current DGX Spark setup into context — models, endpoints, ports, and active containers. Invoke as /spark-status before any task that needs to hit a Spark endpoint. Also use proactively when the user asks about the Spark, mentions a model name, or…
Idioma do texto original: inglês
Generate structured session-summary.md files from campaign chapter PROSE, for chapters that have no session recording behind them — the case where the extraction ladder would otherwise be stuck on raw prose and the event spine has no scene key. Opus…
Idioma do texto original: inglês
Run a consistency check on a session document (first-pass recap, enhanced recap, or final narration) against the campaign's context files, then adjudicate what the docs can't settle against the session transcript. Handles backfilled old chapters, where the…
Idioma do texto original: inglês
Write the full POV-rotating narrative session doc (session-summary-fable-doc.md) in a single pass, directly from scene_extractions_smoothed/ + the voice/ specs + examples/, with the recurring voice-critic findings (em-dash overuse, bookkeeping-noun caps,…
Idioma do texto original: inglês
Set up MemPalace semantic search for a D&D campaign workspace. Use when the user asks to set up mempalace, add mempalace to a campaign, or wants to make a campaign searchable.
Idioma do texto original: inglês
Audit and repair a campaign entity registry that has transcription garblings recorded as aliases, garbled canonical names, or duplicate entities. Enforces the rule that an alias is an APPROVED CANONICAL ALTERNATE NAME, never a misspelling. Every removal,…
Idioma do texto original: inglês
Run scene_extract over a session VTT when the party's PCs are voiced by one or more people — build the voicing map, fix the party.md format gate, pick an attribution strategy at a human checkpoint, run with a sane output ceiling, then verify the output and…
Idioma do texto original: inglês
Re-transcribe a D&D session's Zoom .m4a recording into a more accurate WebVTT, using the campaign's own proper-noun vocabulary while keeping Zoom's existing speaker/timing boundaries. Runs faster-whisper on the DGX Spark via ~/src/mytools/audio-to-vtt. Invoke…
Idioma do texto original: inglês
Merge type-duplicate dossiers in docs/ensemble/state_dossiers/ — the same entity extracted under two or more different (type, subject) keys, e.g. npc_glabbagool.md + monster_glabbagool.md — into docs/ensemble/merged_dossiers/. A shared name across types is…
Idioma do texto original: inglês
Walk the UNKNOWN-surface-form queue emitted by CampaignGenerator's `registry.py triage-candidates` — proper nouns that appear in session outputs but aren't in the entity registry yet. For each candidate the GM decides: new entity, alias of an existing entity,…
Idioma do texto original: inglês
Extract a descriptive proper-noun inventory (NPCs, deities, locations, creatures, items, factions, events) from a module's SOURCE material — a prose bible (.md/.txt) or a 5etools-style adventure JSON — before or alongside running a campaign in it. Produces…
Idioma do texto original: inglês
Review and merge near-duplicate categories produced by drive-tagger's deterministic pipeline. Interactive, one cluster at a time. Invoke as /drive-consolidate.
Idioma do texto original: inglês
Reconcile the live DGX Spark state against current-setup.md. Loads the doc (via spark-status), inspects the actual vllm-chat containers on both boxes, and reports the drift. On approval it conforms the BOXES to the doc (runs the documented spin-up commands);…
Idioma do texto original: inglês
Review undecided alias groups in docs/ensemble/ with campaign-context reasoning — one group at a time, conversationally. Claude looks up each candidate in campaign docs, gives a recommendation with reasoning, and the user confirms or overrides. Decisions…
Idioma do texto original: inglês
Load the four grounding docs (campaign_state, world_state, planning, party) for the current campaign before session prep. Use when the user says "campaign prep", "prep mode", or invokes /campaign-prep.
Idioma do texto original: inglês
Deduplicate NPC dossier files in docs/npcs/ produced by CampaignGenerator's `planning.py --build-dossiers`. Use when the user asks to dedupe dossiers, merge NPCs, or clean up docs/npcs/. Invoke as /dossier-merge [dossier-dir].
Idioma do texto original: inglês
Pre-extraction pass — run enhance_summary then consistency_check against a gm-assist + VTT, before any per-scene VTT extraction. Use when the user invokes /gmassist-precheck [session-dir]. Catches canon and scope problems on the human-authored scene structure…
Idioma do texto original: inglês
Create style example files for session_doc.py narration from the user's campaign summaries. Use when the user asks to create examples or style references for a campaign.
Idioma do texto original: inglês
Create per-character style example files for session_doc.py narration. Use when the user asks for per-character examples, voice examples, or wants to make narrators sound distinct. Invoke as /voice-examples <name1>, <name2>, ...
Idioma do texto original: inglês
Create a per-character voice file for session_doc.py narration. Use when the user asks to create a voice file or character voice notes. Invoke as /voice-file <character_name>.
Idioma do texto original: inglês