Project Telos
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
No arguments | |||
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| telos.statusA | Use when a host needs current Telos workbench readiness and next actions. Read-only, zero-auth, no external side effects. Returns a JSON action envelope. |
| telos.doctorA | Use before demos, listings, or agent runs to check local Telos operator-spine health. Read-only, zero-auth, no external side effects. Returns JSON check results. |
| telos.roomA | Use when an agent needs the current five-flagship room summary before routing work. Read-only and zero-auth. Requires the sibling gather, crucible, index, and forum source checkouts next to this repo plus a local python interpreter; without them it returns an UNVERIFIABLE envelope naming the missing dependency rather than a live summary. Returns a JSON action envelope. |
| telos.workflowA | Use when validating the local five-flagship golden workflow from source checkouts. Read-only and zero-auth, no external side effects beyond local subprocess reads. Requires the sibling gather, crucible, index, and forum source checkouts next to this repo plus a local python interpreter; without them it returns an UNVERIFIABLE envelope naming the missing dependency rather than running the workflow. Returns JSON receipts and verdict counts. |
| telos.catalogA | Use when a host needs the provider-neutral catalog of Project Telos MCP tools and next actions. Read-only, zero-auth, no external side effects. Returns a JSON catalog. |
| telos.server.manifestA | Use when configuring MCP clients for gather, index, forum, crucible, and telos source checkouts. Read-only, zero-auth, no external side effects. Returns a JSON server manifest. |
| telos.mcp.freshnessA | Use when a host must compare loaded MCP servers against expected versions, tools, and probes. Read-only, zero-auth, no external side effects. Returns JSON MATCH, DRIFT, or UNVERIFIABLE freshness receipts. |
| telos.ci.doctorA | Use when a host needs GitHub Actions runtime, action-major, and latest five-flagship CI compatibility receipts. Read-only, zero-auth, no external side effects. Returns a JSON CI doctor register. |
| telos.ci.triageA | Use when a host needs to separate fatal GitHub Actions gate failures from Node runtime migration warnings before routing remediation. Read-only, no external side effects, and no external writes. Supports zero-auth offline fixture packets and live gh run intake through the CLI using local gh auth when required. Returns JSON MATCH, DRIFT, or UNVERIFIABLE CI triage receipts. |
| telos.presentation.doctorA | Use when a host needs five-flagship README, changelog, and brand-asset presentation parity receipts. Read-only, zero-auth, no external side effects. Returns JSON MATCH, DRIFT, or UNVERIFIABLE presentation receipts. |
| telos.accessibility.doctorA | Use when a host needs static HTML accessibility, reduced-motion, keyboard, and canvas-fallback receipts for Telos Studio surfaces. Read-only, zero-auth, no external side effects. Returns JSON MATCH, DRIFT, or UNVERIFIABLE accessibility receipts. |
| telos.performance.doctorA | Use when a host needs static Studio performance, efficiency, asset-budget, and embedding receipts. Read-only, zero-auth, no external side effects. Returns JSON MATCH, DRIFT, or UNVERIFIABLE performance receipts. |
| telos.compatibility.doctorA | Use when a host needs CLI, MCP, protocol, manifest, and integration compatibility receipts. Read-only, zero-auth, no external side effects. Returns JSON MATCH, DRIFT, or UNVERIFIABLE compatibility receipts. |
| telos.operator.doctorA | Use when a host needs README, status, catalog, manifest, CI, and current-state discoverability receipts for the Telos operator surface. Read-only, zero-auth, no external side effects. Returns JSON MATCH, DRIFT, or UNVERIFIABLE operator receipts. |
| telos.admission.telemetryA | Use when designing trace fields that keep action admission separate from verification verdicts. Read-only, zero-auth, no external side effects. Returns a JSON telemetry convention. |
| telos.context.envelopeA | Use when large-workspace context needs readable source refs, budgets, and receipt chains. Read-only, zero-auth, no external side effects. Returns a JSON context-envelope convention. |
| telos.context.packA | Use when preparing a bounded handoff packet for large-codebase agent work. Read-only, zero-auth, no external side effects. Returns a validated JSON context pack with hashes and verdicts. |
| telos.action.receiptA | Use when modeling proposed action, admission, execution, review, and compensation records. Read-only, zero-auth, no external side effects. Returns a JSON action-receipt interface. |
| telos.loop.ledgerA | Use when an agent loop needs durable state across fresh contexts and bounded scheduled runs. Read-only, zero-auth, no external side effects. Returns a JSON loop-ledger convention. |
| telos.objective.monitorA | Use when checking whether proxy metrics are drifting away from real agent or build objectives. Read-only, zero-auth, no external side effects. Returns JSON objective-monitor signals. |
| telos.model.foundryA | Use when planning model-foundry work across hosted, local, open-weight, post-training, and verifier gates. Read-only, zero-auth, no external side effects. Returns a JSON model-foundry contract. |
| telos.learning.forgeA | Use when video, channel, paper, and benchmark leads need to become receipt-backed learning labs before synthesis. Read-only, zero-auth, no external side effects. Returns a JSON Learning Forge packet. |
| telos.learning.labsA | Use when Learning Forge source leads need executable lab contracts with measurements, failure cases, and five-flagship ownership. Read-only, zero-auth, no external side effects. Returns a JSON Learning Forge labs packet. |
| telos.research.seedA | Use when terse research notes need source-backed seed packets before synthesis. Read-only, zero-auth, no external side effects. Returns JSON research seeds with provenance status. |
| telos.research.thermodynamicA | Use when exploring thermodynamic or stochastic AI-chip research through public transcript evidence. Read-only, zero-auth, no external side effects. Returns a JSON research receipt with verification labels. |
| telos.rendering.researchA | Use when collecting rendering leads for clustered-forward, Gaussian splatting, creative coding, and graphics demos. Read-only, zero-auth, no external side effects. Returns JSON research seeds. |
| telos.rendering.capabilitiesA | Use when a host must choose WebGPU, WebGL, canvas, or static rendering fallbacks for Studio surfaces. Read-only, zero-auth, no external side effects. Returns a JSON renderer capability contract. |
| telos.measurement.layersA | Use when visual, splat, lighting, dither, audio, uncertainty, or frame-budget evidence needs meters. Read-only, zero-auth, no external side effects. Returns JSON measurement layers. |
| telos.creative.engineA | Use when presenting the Telos Creative Engine across generative art, sound, typography, media, CGI, math, and physics lanes. Read-only, zero-auth, no external side effects. Returns a JSON creative-engine manifest. |
| telos.creative.kernelsA | Use when deterministic creative primitives are needed for dithering, pixel sorting, plotter paths, or clustered-light bins. Read-only, zero-auth, no external side effects. Returns JSON creative kernels. |
| telos.revival.registryA | Use when deciding which older, siloed, or frozen local tools should be promoted into flagship lanes. Read-only, zero-auth, no external side effects. Returns a JSON revival registry. |
| telos.second_level.queueA | Use when assessing public-safe second-level flagship candidates before registry promotion. Read-only, zero-auth, no external side effects. Returns a JSON second-level flagship queue. |
| telos.workstation.substrateA | Use when a host needs public-safe aggregate intake for local workstation repositories and private/local lane families. Read-only, zero-auth, no external side effects. Returns a JSON workstation substrate register. |
| telos.display.calibrationA | Use when display, color, ICC/LUT, artifact refs, or measurement gates need a non-mutating calibration contract. Read-only, zero-auth, no external side effects. Returns a JSON display-calibration contract. |
| telos.native.controlA | Use when a host needs the Telos native background-control capability catalog for browser (Chrome DevTools Protocol) and native-app (Windows UI Automation) actuation. Read-only, zero-auth, no external side effects. Returns a JSON capability and verb catalog. |
| telos.browser.evidenceA | Use when a host needs the synthetic browser evidence fixture for contract inspection. Read-only, zero-auth, no external side effects. Returns a JSON browser-evidence packet marked unredacted; URL and title context is retained. |
| telos.showcase.scoutA | Use when a host needs fixture-backed OSS Proof Showcase candidate rankings before public patch work. Read-only, zero-auth, no external side effects. Returns JSON scout results. |
| telos.proofB | Use when a host needs the fixture-backed agent-action proof packet joining source refs, context refs, route, admission, side effects, output digests, verifier checks, and the Emet witness stage. Read-only, zero-auth, no external side effects beyond local subprocess reads. Returns a JSON agent-action proof packet. |
| telos.proof.researchA | Use when a host needs the fixture-backed research-claim proof packet joining source provenance, a bounded claim, a required negative control fixture, attempt records, recomputed source digests, and a promotion rung the verifier derives rather than trusts. Read-only, zero-auth, no external side effects beyond local subprocess reads. Returns a JSON research-claim proof packet. |
| telos.proof.visualA | Use when a host needs the fixture-backed visual-truth proof packet whose color and luminance measurements are recomputed from the artifact's own embedded sRGB samples, with a read-only boundary and no physical-calibration overclaim. Read-only, zero-auth, no external side effects beyond local subprocess reads. Returns a JSON visual-truth proof packet. |
| telos.proof.buildA | Use when a host needs the fixture-backed build scientific-runtime proof packet whose conserved-quantity invariant and conservation drift are recomputed from the run's own embedded samples with stdlib math, checked within bounded tolerances, and controlled by a required negative fixture that must break the invariant. Read-only, zero-auth, no external side effects beyond local subprocess reads. Returns a JSON build proof packet. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 41 tools
Most tools have distinct, domain-specific purposes (e.g., ci.doctor vs presentation.doctor, various proof types). However, a few could be confused (catalog vs status vs server.manifest, context.envelope vs context.pack), and the large number increases selection difficulty.
All tools follow a consistent hierarchical dot-separated namespace with lowercase snake_case (e.g., telos.catalog, telos.ci.doctor, telos.proof.visual). The pattern is predictable across all 41 tools.
With 41 tools, the surface is well above the typical 3-15 range and exceeds the 25+ threshold for 'too many.' This large number imposes significant cognitive load and likely includes tools that could be consolidated.
The set covers a wide range of read-only diagnostic and contract areas (CI, presentation, accessibility, proofs, research, etc.), but lacks any mutation or action tools. For a read-only operator spine, coverage is strong, though some gaps may exist.