io.github.travisbergen2/rpcs1-agent-tuner

Est-ce que io.github.travisbergen2/rpcs1-agent-tuner est sûr ?

Au dernier audit MCP Hub, le serveur MCP « io.github.travisbergen2/rpcs1-agent-tuner » est classé « Non encore testé ». Référencé dans l'annuaire. Seul le pentest comportemental fait foi, et il n'a pas encore été exécuté sur ce serveur. Le verdict est réévalué en continu ; consultez la fiche pour l'état à jour.

Méthode : MCP Hub analyse la description de chaque outil (couche 1) et, pour le verdict fort, exécute le serveur dans une micro-VM isolée avec leurres et réseau bloqué (couche 2). On publie des noms et des compteurs, jamais le contenu brut renvoyé par un outil.

Les 9 outils

• calibrate_profile≈ 317 tokens

  Build a ReceiverProfile (TI, SG, FT, UE, AR — continuous 0-100, never a category label) from five behavioral forced-choice answers. Call with NO answers to get the five questions to ask the user; call again with their answers (a/b/c per primitive) to get the profile. Store the returned profile JSON in the user’s notes or memory and pass it to render_reply / prepare_prompt on every turn. Deterministic and stateless — nothing is stored server-side. Schema: https://rpcs1.dev/v1/receiver-profile.json

  Paramètres, retour & exemple (1)

  Paramètres

  • answers · object · optionnel
    Chosen option id per primitive. Omit entirely to receive the questions.

  Exemple d'appel

  {"name": "calibrate_profile", "arguments": {"answers": {}}}

  Exemple de retour

  {"content": [{"type": "text", "text": "…"}]}

• fork≈ 310 tokens

  The calibrated ambiguity surface: deterministic structural fork detectors (reference, scope, grouping, compare-vs-choose, polysemy) with character-offset spans, plus per-reading one-line clarifiers the sender can append to lock a reading in. Returns competing readings, an ask-back question, and a forked-answer scaffold. Silent on clean text by contract. Runs the deterministic mirror floor only over MCP (no model). Prefer this over interpret for span-level ambiguity detection: interpret’s entity list is a word-list engine (calibrated 2026-08-15: no discrimination on conversational text) — advisory only.

  Paramètres, retour & exemple (2)

  Paramètres

  • text · string · requis
    The message to analyze for forks.
  • rejected · array · optionnel
    Reading summaries the user already rejected — never re-offered.

  Exemple d'appel

  {"name": "fork", "arguments": {"text": "texte"}}

  Exemple de retour

  {"content": [{"type": "text", "text": "…"}]}

• interpret≈ 229 tokens

  Detect ambiguity in user messages using the RPCS-1 Signature Ambiguity Framework. Returns AR level (AR0-AR5), confidence, candidate interpretations with scores, clarifying questions, and suggested next step. Use when a user says something vague, passive-aggressive, or underspecified.

  Paramètres, retour & exemple (2)

  Paramètres

  • risk · string · optionnel · casual | advice | high-stakes | safety-critical
    Risk category for ambiguity threshold.
  • text · string · requis
    The message to interpret.

  Exemple d'appel

  {"name": "interpret", "arguments": {"text": "texte"}}

  Exemple de retour

  {"content": [{"type": "text", "text": "…"}]}

• normalize≈ 154 tokens

  Clean up text with ellipses, fragments, and run-on thoughts into coherent prose. Use when a user types stream-of-consciousness or fragmented input.

  Paramètres, retour & exemple (1)

  Paramètres

  • text · string · requis
    Fragmented text to normalize.

  Exemple d'appel

  {"name": "normalize", "arguments": {"text": "texte"}}

  Exemple de retour

  {"content": [{"type": "text", "text": "…"}]}

• prepare_prompt≈ 541 tokens

  The inbound half of the Translation Bridge loop. Takes the user’s raw message (possibly ambiguous, fragmented, or underspecified) plus their ReceiverProfile, and returns the recovered intent, a canonical translation to act on, ambiguity level, and — profile-aware — whether to clarify or commit. Call this before acting on any ambiguous user request. Scope note: its detectors are lexical/structural (vague signals, ambiguous references) — for the commit-vs-clarify DECISION, route_intent (with your own proposed readings) is the authority; when they disagree, follow route_intent.

  Paramètres, retour & exemple (3)

  Paramètres

  • risk · string · optionnel · casual | advice | high-stakes | safety-critical
    Risk category for the ambiguity threshold.
  • text · string · requis
    The user’s raw message.
  • profile · object · optionnel
    The user’s ReceiverProfile from calibrate_profile. Shapes clarify-vs-commit behavior.

  Exemple d'appel

  {"name": "prepare_prompt", "arguments": {"text": "texte"}}

  Exemple de retour

  {"content": [{"type": "text", "text": "…"}]}

• recommend_agent_configuration≈ 1620 tokens

  Diagnose why a deployed AI agent may fail. Takes environmental entropy, predictability, stakes, context horizon, and commitment style, then returns receiver profile values (TI, SG, FT, UE, AR), platform parameters (temperature, top_p, strategy), regime prediction, reasoning, and warnings. Optionally pass target_model (the actual model id) to attach MEASURED per-model receiver posture (E-LIT table): evidence-graded literalness, truth-override boundary, and translation directives. Deterministic, stateless, read-only — does not store past recommendations.

  Paramètres, retour & exemple (4)

  Paramètres

  • task · object · optionnel
  • environment · object · optionnel
  • target_model · string · optionnel
    Optional: the actual model id this agent will run on (e.g. "claude-sonnet-4-6", "deepseek-v4-pro"). When it matches a measured per-model receiver entry (E-LIT table), measured translation directives and evidence-graded posture data are attached to platform_parameters. Unknown models fall back to platform-level behavior unchanged.
  • target_platform · string · optionnel · anthropic | openai | open_source | generic
    The platform whose runtime parameters should be recommended.

  Exemple d'appel

  {"name": "recommend_agent_configuration", "arguments": {"task": {}, "environment": {}, "target_model": "texte", "target_platform": "anthropic"}}

  Exemple de retour

  {"content": [{"type": "text", "text": "…"}], "structuredContent": {"warnings": [], "reasoning": "texte", "confidence": "high", "predicted_regime": "stable", "receiver_profile": {}, "platform_parameters": {}, "imm_principles_applied": []}}

• render_reply≈ 435 tokens

  The outbound half of the Translation Bridge loop. Takes your draft reply plus the user’s ReceiverProfile and returns deterministic rendering instructions (structure, warmth, explicitness, revision posture, ambiguity handling — each with a why-trace). Apply the instructions to your draft before answering. Call this on every reply to a calibrated user.

  Paramètres, retour & exemple (2)

  Paramètres

  • text · string · requis
    Your draft reply.
  • profile · object · requis
    The user’s ReceiverProfile from calibrate_profile.

  Exemple d'appel

  {"name": "render_reply", "arguments": {"text": "texte", "profile": {}}}

  Exemple de retour

  {"content": [{"type": "text", "text": "…"}]}

• rewrite≈ 200 tokens

  Get rewrite instructions for adapting text to a specific style: technical, plain, socially_gentle, concise, detailed, or direct. Use when communication needs tone adjustment.

  Paramètres, retour & exemple (2)

  Paramètres

  • text · string · requis
    Text to rewrite.
  • style · string · optionnel · technical | plain | socially_gentle | concise | detailed | direct
    Target audience style.

  Exemple d'appel

  {"name": "rewrite", "arguments": {"text": "texte"}}

  Exemple de retour

  {"content": [{"type": "text", "text": "…"}]}

• route_intent≈ 968 tokens

  Entropy routing over competing interpretations — the model proposes, the deterministic core disposes. YOU generate the candidate readings of the user’s message (3–7 short hypotheses covering the plausible interpretations, INCLUDING likely-typo readings, idiom-vs-literal readings, and domain senses) and pass them as hypotheses, ideally with your own likelihoods (0–1 per reading) AND a paraphrase per reading — the user’s message rewritten unambiguously under that interpretation, so the user can VERIFY intent by recognition before anything commits (one misread prompt skews a whole thread). The router computes the posterior and its normalized entropy T̂ and returns the decision: commit (one reading dominates), commit_with_note (close alternative disclosed), present_options (several readings live), or clarify (ask before acting — open-endedly when nothing discriminates). Thresholds adapt to the user’s ReceiverProfile (AR widens/narrows the commit region; high FT discloses near-ties). This tool is the commit-vs-clarify AUTHORITY in the pipeline. Omitting hypotheses falls back to a generic six-intent PRODUCT-ROUTING starter set — do not use the fallback for interpreting arbitrary sentences. Deterministic, stateless, read-only. Benchmarked: RTEB v1.1 (developer-bench grade; see docs/routing.md).

  Paramètres, retour & exemple (4)

  Paramètres

  • text · string · requis
    The user’s raw message.
  • profile · object · optionnel
    The user’s ReceiverProfile from calibrate_profile. Shapes commit-vs-clarify thresholds.
  • hypotheses · array · optionnel
    Candidate interpretations. Omit to use a generic six-intent starter set plus a catch-all.
  • likelihoods · object · optionnel
    Optional externally computed likelihood per hypothesis id (e.g. model-derived) — replaces the lexical scorer.

  Exemple d'appel

  {"name": "route_intent", "arguments": {"text": "texte"}}

  Exemple de retour

  {"content": [{"type": "text", "text": "…"}]}

C'est votre serveur ?

Revendiquez-le sur l'espace créateur (preuve GitHub, domaine ou npm) pour afficher le badge « propriétaire vérifié » et lancer un pentest comportemental à la demande afin d'obtenir le badge « Vérifié ».

Créer un compte / revendiquer →

Source

https://rpcs1.dev/mcp