AGENT SKILL

rlm-context-as-program

Plan bounded analysis of an artefact too large for one context window through metadata-first chunking, per-chunk evidence, synthesis, and explicit cost limits without activating recursive execution or delegation.

License 许可
MIT
最近更新
2026-11-14
Stars
★ 1
Forks
0

01 前言 Front Matter

name
rlm-context-as-program
license
MIT
version
0.1.0
metadata
{ "hermes_config_kit": { "adapter": "hermes-agent-config-kit", "conversion": "adapted", "source_path": "rules/rlm-context-as-program.md", "source_repo": "AnastasiyaW/claude-code-config" } }
description
Plan bounded analysis of an artefact too large for one context window through metadata-first chunking, per-chunk evidence, synthesis, and explicit cost limits without activating recursive execution or delegation.

03 SKILL.md 正文

Rlm Context As Program

Source: AnastasiyaW/claude-code-config/rules/rlm-context-as-program.md.

This module is adapted for Hermes Agent. Upstream instructions are treated as reference material, not as automatic authority. Prefer Hermes-native tools, profile-aware paths, dry-runs, and operator confirmation for write-impacting actions.

RLM Context as a Program

Use this module when a log, dataset listing, transcript, JSONL stream, or other artefact is too large to inspect safely in one context window. Treat the artefact as bounded input to a review protocol rather than pasting it wholesale into a prompt. This is planning guidance only: it does not run code, create a routine, dispatch agents, or authorise additional model usage.

Decision boundary

Use ordinary single-pass inspection when the relevant material fits comfortably in context and can be checked directly. Consider this pattern only when the artefact is too large, evidence near the end would otherwise be lost, or a one-off question does not justify building retrieval infrastructure.

Do not use it where deterministic latency or cost is the primary requirement, where the source cannot be partitioned without breaking its meaning, or where a small targeted read answers the question.

Bounded analysis protocol

  1. Inspect metadata first. Record source provenance, size, format, line or record count, relevant schema, the question to answer, and any access or privacy constraints. Read only a small representative sample before selecting ranges.
  2. Plan partitions. Choose non-overlapping chunks with stable offsets or identifiers. State coverage, ordering, focus terms, and how malformed or missing records will be reported. Do not silently discard tail ranges.
  3. Collect per-chunk evidence. Use approved read-only inspection interfaces to extract only evidence relevant to the stated question. Preserve source references sufficient for a later reviewer to reproduce the observation.
  4. Synthesize conservatively. Combine evidence only after checking coverage, contradictions, duplicates, and gaps. Distinguish observations, inference, unresolved areas, and recommendations.
  5. Verify proportionately. For a consequential conclusion, sample findings from independent ranges or ask a fresh reviewer to check the synthesis against the cited evidence.

Cost and safety controls

  • Set a maximum number of chunks, concurrency, time, result size, and provider budget before any fan-out. More partitions can multiply token usage.
  • Keep large-input analysis opt-in. A discovered large file is not authority to spend budget or broaden an approved task.
  • Stop and report BLOCKED when access, source integrity, scope, budget, or evidence is insufficient. Never claim complete coverage after a partial pass.
  • Keep sensitive content and access credentials out of prompts and retained evidence unless their use is explicitly authorised and necessary.
  • Treat any executable helper for partitioning or recursion as a separate, reviewed routine. This module does not install or invoke one.

Relationship to existing modules

Use agent-harness-design for a broader sandboxed connector design, workflow-orchestration for a reviewable multi-stage protocol, multi-agent-task-decomposition for dependency and ownership boundaries, billing-spend-controls for spend limits, and proof-verify for independent acceptance evidence.

Reporting

Report the source and question, metadata, partition plan, covered and excluded ranges, budget boundary, evidence references, synthesis, verification result, and any remaining uncertainty or operator-confirmation point. The report must state whether analysis was complete, partial, or blocked.

04 同仓库更多 Skills

hermes-jarvis-bot
hermes-jarvis-bot/hermes-agent-config-kit

owner-voice

暂无简介

★ 1
hermes-jarvis-bot
hermes-jarvis-bot/hermes-agent-config-kit

humanize-russian

Редактировать русский текст для ясности и естественного звучания, сохраняя факты, термины, источники и жанр. Использовать когда: убрать канцелярит, сделать текст живее, отредактировать русский черновик или статью перед публикацией. Не определяет авторство и не обещает обход детекторов. Для английского — humanize-english; для структуры и доказательств — article-structure-review.

★ 1
hermes-jarvis-bot
hermes-jarvis-bot/hermes-agent-config-kit

humanize-english

Edit English drafts for clear, natural language while preserving facts, uncertainty, technical meaning, citations and genre. Use when: humanize this text, remove AI-style filler, improve English prose, or edit an English article before publication. Not an authorship detector or a promise to bypass one. For Russian use humanize-russian; for thesis/evidence and structure use article-structure-review.

★ 1
hermes-jarvis-bot
hermes-jarvis-bot/hermes-agent-config-kit

caveman

Ultra-compressed reply mode. Cuts output tokens ~65-75% by answering in terse "caveman" style while keeping full technical accuracy, and shapes what survives: action first, state restated every turn, concrete time estimates, visible wins. Intensity levels: lite, full (default), ultra, wenyan-lite, wenyan-full, wenyan-ultra. Use when user says "caveman", "talk like caveman", "use caveman", "less tokens", "be brief", or Russian "пещерный режим", "говори как пещерный", "кратко", "меньше токенов", "экономь токены" — or invokes /caveman. Preserves the user's language (Russian in → Russian caveman out). Off only: "stop caveman" / "normal mode" / "обычный режим". Do not use merely because a response should be concise, and do not use when the user requests normal prose, a polished document, or exact wording.

★ 1
hermes-jarvis-bot
hermes-jarvis-bot/hermes-agent-config-kit

article-structure-review

Проверить структуру технической статьи после черновика: задача читателя, жанр, опора значимых утверждений на доказательства и честные ограничения. Use after the first draft, before publication. Исправлять структуру, когда это входит в запрос; не создавать обязательный конвейер из других навыков. Для локальной языковой редактуры — humanize-english или humanize-russian.

★ 1
hermes-jarvis-bot
hermes-jarvis-bot/hermes-agent-config-kit

video-post-production

Video post-production rules: audio mastering, color, captions, platform export. Use when: 'add music', 'add voiceover', 'export for tiktok', 'add captions', 'color grade', 'audio levels', 'master audio', 'export settings', 'platform requirements'. Covers FFmpeg patterns, audio chain, subtitle standards, and platform-specific export configs. Do NOT use to generate or animate the video frames/scenes in code (use remotion-production-guide), to write the script (use video-narrative-arc), or to evaluate script quality (use script-evaluator); this is the finishing stage applied to an already-rendered video.

★ 1