PlatPhorm Knowledge Hub
Turn public content and tool output into attributed, searchable, agent-readable documentation.
Ingest URLs, feeds, Markdown, public documents, reports, podcasts, specs, tool outputs, and workflow evidence. OpenDocs extracts, attributes, ranks, scores, tags, links, and publishes content across the PlatPhormNews network.
Documents
77
Published documentation pages
Live server count
Submissions
103
Content submissions received
Live server count
Sources
65
Unique content sources
Live server count
What OpenDocs Does
OpenDocs is the canonical public documentation, ingestion, attribution, ranking, scoring, search, clipping, publishing, MCP/API docs, agent-readable knowledge, and evidence-report layer for PlatPhormNews.
Ingest
Fetch, parse, clean, and preview public-safe sources with SSRF guards and explicit degraded states.
Attribute
Keep original source URL, canonical URL, source domain, author/date when found, citation text, and hashes.
Score
Explain quality, freshness, source confidence, uniqueness, safety, and agent-readiness without fake popularity.
Publish
Expose human pages, API, MCP resources, llms files, RSS, sitemap, policy, and CLI workflows.
Ingestion Sources
8 source types have public-safe preview support. Private connectors and durable imports degrade or require PLATPHORM_API_KEY.
Integrations
3 integrations are currently available in this checkout and 14 are honest degraded/protected connectors.
Attribution and Trust
Mandatory source-backed attribution with unknown fields labeled, not invented.
Ranking and Scoring
Deterministic scoring explains source credibility, freshness, completeness, duplicate risk, and review priority.
Search and Discovery
Public search, categories, tags, sources, RSS, sitemap, llms, OpenAPI, and MCP discovery.
Reports From Tools
Docs is the target for Evals, BrowserOps, Sandbox, Webhooks, Monitor, Trace, AgentUI, Claws, and platphormctl evidence.
MCP / API / CLI
Public-safe workflows are open by default. Protected persistence, review, publishing, webhook, sync, reporting, and admin actions use PLATPHORM_API_KEY.
Use Docs with an Agent
Read docs, search over MCP, inspect resources, get prompts, and call bounded public tools through /api/mcp.
Clip Content with the API
Preview public sources through /api/v1/clip/preview, then submit cleaned Markdown for review through /api/v1/clip/commit.
Submit Public Content
Send public articles, reports, platform posts, or cleaned clips into the review queue without storing private keys in the browser.
Search Docs over MCP
Use search_docs for published documentation and docs://sources or docs://tags for public attribution and taxonomy context.
Trace Your Ingest Job
Clipping and ingest operations create W3C trace ids and propagate safe PlatPhorm headers to trusted outbound calls.
Public vs Protected
Browsing, search, clipping previews, submissions, discovery, and read-only MCP are public-safe; protected writes require PLATPHORM_API_KEY.
Latent-Space Fine-Tuning
Definition of LoRA, adapters, AWS SageMaker, and Bedrock as latent-space adaptation workflows.
aiplatphorm-dictionary-latent-space-fine-tuning-flat
Latent Space Surgery
Definition of targeted model editing through concept directions in latent space.
aiplatphorm-dictionary-latent-space-surgery-flat
Latent Operations
Definition of vector shifts, interpolation, slicing, masking, and sampling in latent space.
aiplatphorm-dictionary-latent-operations-flat
Latent Reasoning
Definition of latent reasoning in LLM hidden states and continuous representations.
aiplatphorm-dictionary-latent-reasoning-flat
Embedding Space
Definition of embedding space as a vector representation for semantic similarity and retrieval.
aiplatphorm-dictionary-embedding-space-flat
Latent Space
Definition of latent space in machine learning, LLMs, and embeddings.
aiplatphorm-dictionary-latent-space-flat
Recently Ingested
Review-backed only
Ingestion runs are shown after they produce stored public-safe submissions or documents. No fake activity is displayed.
Highest Confidence
Source-scored docs
Confidence ranking is computed from attribution evidence and appears when scored public documents are available.
Needs Review
Protected queue
Review counts are public-safe summaries; publish and reject actions require PLATPHORM_API_KEY.
Quick Start: Connect Your AI Agent
Use the MCP endpoint to connect any compatible AI agent