Install
$ agentstack add skill-tiangong-ai-agent-skills-bluesky-cascade-fetch ✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.
Security review
✓ PassedNo issues found. Passed automated security review. · v0.1.0 How review works →
- ✓ Prompt-injection patterns
- ✓ Secret / credential exfiltration
- ✓ Dangerous shell & filesystem operations
- ✓ Untrusted network calls
- ✓ Known-malicious package signatures
What it can access
- ✓ Network access No
- ✓ Filesystem access No
- ✓ Shell / process execution No
- ✓ Environment & secrets No
- ✓ Dynamic code execution No
From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.
Verified badge
Passed review? Show it. Paste this badge into your README, it links to the public security report.
Reliability & compatibility
Declared compatibility
Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.
We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.
How agent discovery & health will work →About
Bluesky Cascade Fetch
Use the CLI-owned bluesky.public-posts/fetch-cascades operation. This Skill supplies intent routing and result-use boundaries only; the CLI TypeScript 7 runtime owns source discovery, schemas, endpoint policy, pagination, retries, normalization, limits, partial results, and receipts.
Before running
- Read
references/tiangong-data-requirement.json. - Use the caller- or workspace-resolved stable CLI. The requirement declares
compatible capability and operation contract majors; it does not select a package build.
- Run
data describewith that same CLI. Continue only when the capability
ID and required contract majors match, and copy the exact current capability/operation versions from that response into the run request.
tiangong-ai data describe bluesky.public-posts --json
Use the current Discovery Metadata to confirm source coverage, freshness, restrictions, selection hints, provides, and doesNotProvide. Do not rely on provider facts remembered from an older Skill revision.
Choose the seed source
- Use
searchwhen the user supplied a topic or query. - Use
author-feedonly for a named public actor view. - Use
feedonly when the caller supplied a known public feed-generator AT-URI. - Use
list-feedonly when the caller supplied a known public list AT-URI. - Enable thread expansion only when visible reply topology is needed. Prefer
seed-only retrieval when discovery breadth matters more than cascades.
- Preserve an explicit user time window and filters. Never widen or replace
them silently.
- When diagnosing an empty historical search, repeat the same bounded request
with source.applyServerTimeFilter set to false. Keep the client-side time window unchanged and classify any still-empty result as provider coverage, not proof that no discussion occurred.
Prepare the request
Build one tiangong.data.run-request.v1 envelope and validate it against the current input schema from data describe. Replace both version placeholders with the exact versions from the same data describe response.
{
"schemaVersion": "tiangong.data.run-request.v1",
"capabilityId": "bluesky.public-posts",
"capabilityVersion": "",
"operationId": "fetch-cascades",
"operationVersion": "",
"input": {
"source": {
"mode": "search",
"query": "climate policy",
"sort": "latest",
"language": "en",
"tags": ["climate"],
"applyServerTimeFilter": true
},
"startDateTime": "2026-03-10T00:00:00Z",
"endDateTime": "2026-03-11T00:00:00Z",
"pageSize": 50,
"expandThreads": true,
"maxThreads": 20,
"threadDepth": 8,
"threadParentHeight": 5
}
}
Do not put an endpoint override, credential, output path, scheduler, or provider-specific field absent from the schema into the request. Persistence and recurring collection belong to the caller.
Run
tiangong-ai data run bluesky.public-posts fetch-cascades \
--input /absolute/path/to/request.json --json
Preserve the complete tiangong.data.run-result.v1 envelope, including contract, warnings, errors, summary, and receipt, when handing results to another workflow.
Result boundaries
- Treat search ranking, feed selection, visibility, timestamps, counters, and
thread nodes as mutable provider snapshots, not exhaustive or verified facts.
- Surface blocked/not-found nodes, failed threads, empty results,
partial, and
truncation. Never reinterpret them as complete absence or complete coverage.
- Retain
hitsTotal, per-page invalid-record counts, cascade validation, and
the fallback-host indicator; they are acquisition diagnostics, not evidence quality scores.
- Public posts can contain personal, sensitive, misleading, or unsafe content.
Quote or retain only what the task requires and preserve provenance.
- Reply topology is a visible snapshot. Do not infer causation, influence,
representativeness, identity, or sentiment labels from it alone.
- Use a separately governed repository/firehose workflow for exhaustive AT
Protocol records and a content workflow for media or linked pages.
- Cross-source comparison, evidence admission, persistence, polling, and
research conclusions belong to the caller or Auto Research.
Reference
references/tiangong-data-requirement.json: stable capability requirement; it is not a package lock.
Source & license
This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.
- Author: tiangong-ai
- Source: tiangong-ai/agent-skills
- License: MIT
Install and usage instructions live in the source repository linked above.
Reviews
No reviews yet, be the first.
Write a review
Versions
- v0.1.0 Imported from the upstream source.