Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add SkardiLabs/skardi-skills --skill auto-contextgit clone --depth 1 https://github.com/SkardiLabs/skardi-skillsWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/skardilabs/skardi-skills/auto-context)<a href="https://agentmods.dev/skills/skardilabs/skardi-skills/auto-context"><img src="https://agentmods.dev/badge/skills/skardilabs/skardi-skills/auto-context/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/skardilabs/skardi-skills/auto-context"><img src="https://agentmods.dev/badge/skills/skardilabs/skardi-skills/auto-context.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00431 | $0.12017 |
| Opus 5 | $0.00216 | $0.06008 |
| Sonnet 5 | $0.00086 | $0.02403 |
| Haiku 4.5 | $0.00043 | $0.01202 |
Grade A, and why
auto-context scanned grade A with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
curl -X POST http://localhost:8080/search-hybrid/execute \ How it starts
The opening of the file, as written. The whole thing — 363 lines — stays where its author put it; the contents beside it link to each section on GitHub.
auto_context — build searchable context an agent can query
Your job: turn a corpus into a working retrieval surface, then answer questions from it. One flow; three ways raw material comes in, two places the index can live.
A server is not optional. Since the CLI was reframed as a thin HTTP client (skardi PR #170) it holds no query engine, no data-source registration, and no local execution mode. Every path in this skill starts a
skardi-server. Do not look for a CLI-only shortcut — there isn't one, and earlier versions of this skill that promised "no server, no Docker" are obsolete.
What must be new enough is the server, not the CLI. Ingest and search each do their work inside one server-side SQL statement: chunking via
chunk(), embedding viacandle()/gguf()/remote_embed(), and writing — one INSERT for ingest, inline embedding for vector and hybrid search. All of those UDFs are registered byskardi-server, so the only thing that matters is which features it was built with. The image that bundles them isghcr.io/skardilabs/skardi/skardi-server-rag:0.5.0(--features rag); the olderskardi-server-embeddingimage does not registerchunk()and breaks the rendered pipelines.Pin the tag. Use
:0.5.0, the current released version — not:latest, which moves under you, and not a build frommain, which carries unreleased changes you cannot name in a bug report. This skill does not check theskardiCLI at all: it is a thin HTTP client, it is never invoked by these scripts, and its version says nothing about the server's UDFs.
Where the raw material comes from — three entries, one flow
Raw material and index storage are independent choices, and they read confusingly alike because both can involve "a database". This section is about where the source text comes from; the next section is about where the index lives. In particular: the "bring your own datastore" storage path stores the index in a new table created for that purpose — it never reads rows you already have. Reading rows you already have is this section's second entry, and it is read-only.
What ships with it
28 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
- assets/postgres/ctx.yaml.tpl 851 B
- assets/postgres/pipelines/ingest_chunked.yaml.tpl 3.0 KB
- assets/postgres/pipelines/ingest.yaml.tpl 1.1 KB
- assets/postgres/pipelines/search_fulltext.yaml.tpl 697 B
- assets/postgres/pipelines/search_hybrid.yaml.tpl 1.7 KB
- assets/postgres/pipelines/search_vector.yaml.tpl 853 B
- assets/postgres/semantics.yaml.tpl 2.0 KB
- assets/sqlite/ctx.yaml.tpl 697 B
- assets/sqlite/pipelines/ingest_chunked.yaml.tpl 3.6 KB
- assets/sqlite/pipelines/ingest.yaml.tpl 1.1 KB
- assets/sqlite/pipelines/search_fulltext.yaml.tpl 1.8 KB
- assets/sqlite/pipelines/search_hybrid.yaml.tpl 1.8 KB
- assets/sqlite/pipelines/search_vector.yaml.tpl 763 B
- assets/sqlite/semantics.yaml.tpl 2.0 KB
- references/fetch_and_land.md 13 KB
- references/pipeline_patterns.md 9.7 KB
- references/runtimes.md 9.5 KB
- references/schemas.md 6.3 KB
- references/troubleshooting_sqlite.md 14 KB
- references/troubleshooting.md 13 KB
- scripts/_diagnose.py 8.4 KB runs code
- scripts/_platform.py 2.5 KB runs code
- scripts/_report.py 6.0 KB runs code
- scripts/ingest_corpus.py 44 KB runs code
- scripts/ingest_table.py 29 KB runs code
- scripts/setup_context.py 32 KB runs code
- scripts/start_server.py 51 KB runs code
- scripts/stop_server.py 5.8 KB runs code
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2d ago Changed · +2 lines f509c0c997b8
- 7d ago Changed · +12 lines 45e91c9b547a
- 11d ago First seen · 349 lines · 431 tokens per session scan A 27cfd213cf07
auto-context is a skill published in the GitHub repository SkardiLabs/skardi-skills (5 stars, last pushed yesterday), licensed Apache-2.0. It adds 431 tokens to every session and 12,017 once invoked, about $0.0022 per session on Opus 5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
AgentDB Memory Patterns
Implement persistent memory patterns for AI agents using AgentDB. Includes session memory, long-term storage, pattern learning, and context management. Use when building stateful agents, chat systems, or intelligent assistants.
pinecone-research
Agent RAG and long-term memory with Pinecone.
agent-v3-memory-specialist
Agent skill for v3-memory-specialist - invoke with $agent-v3-memory-specialist.
langchain
Framework for building LLM-powered applications with agents, chains, and RAG. Supports multiple providers (OpenAI, Anthropic, Google), 500+ integrations, ReAct agents, tool calling, memory management, and vector store retrieval. Use for building chatbots, question-answering systems, autonomous agents, or RAG…
browserwing-admin
Manage and operate BrowserWing — an intelligent browser automation platform. Install dependencies, configure LLM, create/manage/execute automation scripts, use AI-driven exploration to generate scripts, browse the script marketplace, and troubleshoot issues.
mem0-integration
Mem0 memory layer integration for AI agents. Implement persistent, semantic memory for long-term context retention and personalization.