Brief
LlamaIndex 0.14.15 adds an AgentMesh trust layer and retry handling for extractors
Run-llama published llama-index v0.14.15 on 2026-02-18, adding a new agent trust-layer package and retry and error handling in BaseExtractor. The release notes are one-line entries, so the operational detail behind each item is not stated.
Two entries in the llama-index v0.14.15 notes describe changes a workflow operator could act on. The first is a new package, llama-index-agent-agentmesh 0.1.0, described only as "AgentMesh: Trust Layer for LlamaIndex Agents" (#20644). The second is in core: retry and error handling added to BaseExtractor (#20693), the component underneath extraction steps in indexing and retrieval pipelines.
Alongside those, llama-index-program-evaporate 0.5.2 sandboxes LLM-generated code execution in EvaporateExtractor (#20676), and llama-index-observability-otel 0.3.0 improves OpenTelemetry data serialization by flattening dicts (#20719).
Our reading
For this desk, the BaseExtractor change is the one with a clear operational footprint: extraction is a step teams already run inside indexing pipelines, so added retry and error handling changes failure behaviour at that step. The AgentMesh package is a new integration point for teams putting agents behind some trust control, but the notes give only its name, so what it enforces is unknown. The O…
What to do or watch
Before upgrading, pin the version and re-run any pipeline whose extraction or observability output you depend on, checking that flattened OpenTelemetry attributes and new retry behaviour do not break downstream parsing, dashboards or alert rules. The precise unresolved question is what the AgentMesh trust layer actually enforces.
Source details and supporting facts
Each line is stated by the page named above it.
Stated by github.com
- The release notes are dated 2026-02-18 and cover llama-index v0.14.15.
- llama-index-agent-agentmesh 0.1.0 is listed as "AgentMesh: Trust Layer for LlamaIndex Agents" (#20644).
- llama-index-core 0.14.15 adds "retry and error handling to BaseExtractor" (#20693).
- llama-index-observability-otel 0.3.0 improves OpenTelemetry data serialization by flattening dicts (#20719).
- llama-index-program-evaporate 0.5.2 sandboxes LLM-generated code execution in EvaporateExtractor (#20676).
- llama-index-readers-github 0.10.0 enhances GitHubRepoReader with selective file fetching and deduplication.
Sources
- LlamaIndex releasesText stored 17 September 2026
How this story was checked. Written from the 1 page listed above, stored 17 September 2026; claims checked against that stored text on 17 September 2026.
What that means
- 6 of 6 reported statements were confirmed against the page that carries them; the rest were removed rather than published.
- Figures in the text were required to appear in the stored source text: yes. Identifiers: yes.
- The check reads stored text only: no claim rests on a fresh look that did not happen.
- Where the reporting was silent, the text says so instead of filling the gap.