Release v0.20.8

[lchoquel]

Release v0.20.8

Bumps version from 0.20.7 to 0.20.8.

Changelog

• Add needs_model_specs=True to agent CLI run and models commands.

---

Note

Medium Risk
Moderate risk: tweaks agent CLI initialization flags to force or conditionally fetch model specs, which could change runtime behavior and trigger additional remote config/model lookups during pipelex-agent commands.

Overview
Cuts release v0.20.8 by bumping pyproject.toml/uv.lock versions and adding a v0.20.8 entry to CHANGELOG.md.

Agent CLI behavior change: pipelex-agent run commands now initialize Pipelex with needs_model_specs=True, and models --backend conditionally enables needs_model_specs so backend-filtering can resolve real model specs even when needs_inference=False.

Updates the internal release skill doc to remove the [Unreleased] changelog workflow (version entries only), and regenerates uv.lock (notably adding platform markers to some deps).

^{Written by Cursor Bugbot for commit 2bd34cc. This will update automatically on new commits. Configure here.}

[cubic-dev-ai]

1 issue found across 7 files

Prompt for AI agents (unresolved issues)

Check if these issues are valid — if so, understand the root cause of each and fix them. If appropriate, use sub-agents to investigate and fix each issue separately.


<file name="pipelex/cli/agent_cli/commands/models_cmd.py">

<violation number="1" location="pipelex/cli/agent_cli/commands/models_cmd.py:243">
P2: Only request model specs when `--backend` is used. As written, `pipelex-agent models` now depends on gateway terms/network even for the default local-only listing path.</violation>
</file>

_{Reply with feedback, questions, or to request a fix. Tag @cubic-dev-ai to re-run a review.}