upstream-watch

upstream-watch is a command for Claude Code from DougTrajano/pydantic-ai-skills. It costs 31 tokens per session (2,561 once invoked), scanned A, original, MIT.

A scheduled review routine for checking new releases of Pydantic AI and its harness. It records whether those releases could affect this package.

In plain words
What is it for?
Use it weekly or before releasing to review new upstream versions and record a verdict in `.github/upstream-watch.md`, along with any issue or pull request needed.
Why use it?
This package relies on internal parts of Pydantic AI, so an upstream release could cause breakage even when this package has not changed. The routine makes that risk visible before a release.

Command for Claude Code

Written for Claude Code: allowed-tools in frontmatter. Also seen: mentions AGENTS.md.

Good fit Use it weekly or before releasing to review new upstream versions and record a verdict in .github/upstream-watch.md, along with any issue or pull request needed.

Compare 6 commands from other repositories ↓
Install with agentmods
npx agentmods add commands/dougtrajano/pydantic-ai-skills/upstream-watch
Install

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.

Clone the repo
git clone --depth 1 https://github.com/DougTrajano/pydantic-ai-skills

Made for: Claude Code.

Wrote 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.

agentmods badge for upstream-watch

README.md
[![agentmods](https://agentmods.dev/badge/commands/dougtrajano/pydantic-ai-skills/upstream-watch/github.svg)](https://agentmods.dev/commands/dougtrajano/pydantic-ai-skills/upstream-watch)
Your own site
<a href="https://agentmods.dev/commands/dougtrajano/pydantic-ai-skills/upstream-watch"><img src="https://agentmods.dev/badge/commands/dougtrajano/pydantic-ai-skills/upstream-watch/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.

agentmods 80×15 button for upstream-watch

Your own site · 80×15
<a href="https://agentmods.dev/commands/dougtrajano/pydantic-ai-skills/upstream-watch"><img src="https://agentmods.dev/badge/commands/dougtrajano/pydantic-ai-skills/upstream-watch.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 31 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 2,561 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5.1 $0.00031 $0.02561
Opus 5 $0.00015 $0.01281
Sonnet 5 $0.00006 $0.00512
Haiku 4.5 $0.00003 $0.00256

Measured 5d ago against content hash 07891af566b0, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-11, from the pricing page.

Security

Grade A, and why

upstream-watch scanned grade A with 0 findings 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 5d 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.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

.claude/commands/upstream-watch.md · 192 lines

How it starts

The opening of the file, as written. The whole thing — 192 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Upstream watch

Two upstream projects can break this package without any change here:

  • pydantic-ai — it wraps the framework and imports several of its private symbols, which no deprecation policy protects.
  • pydantic-ai-harness — a required runtime dependency since v2. SkillsCapability is built directly on its public Skills class, and harness is on 0.x releases where its own README says the API may change between minor releases.

This routine is the standing review that catches both. It produces one entry in .github/upstream-watch.md and, when action is needed, the issue or PR that fixes it.

Run it weekly, or on demand before a release.

1. Establish the lower bound

Read the top entry of .github/upstream-watch.md. Its two "checked through" tags are the lower bound for this run — you are reviewing everything published after them. Note today's date for the new entry heading.

2. Collect the releases

gh release list --repo pydantic/pydantic-ai --limit 40
gh release list --repo pydantic/pydantic-ai-harness --limit 20

Everything newer than the lower bound is in scope for both repositories. The pydantic-ai v1.x backport line is not: pyproject.toml declares pydantic-ai-slim>=2.38, so v1 is no longer a supported configuration. (Entries before 2026-09 reviewed it because the floor was then >=1.105 — that is why the log mentions it.)

Read each release body with gh release view <tag> --repo <repo>. When a release note is ambiguous about what a PR actually touched, open the PR's file list rather than guessing — past entries in the log did exactly this, and it is what turned a scary-sounding release note into a confident "no action".

3. Assess impact against what this package actually uses

Two narrow surfaces. Check each release against them specifically.

3a. pydantic-ai private imports

Read the full file on GitHub · 192 lines

Changes

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.

  1. 5d ago Changed · +95 lines 07891af566b0
  2. 11d ago First seen · 97 lines · 31 tokens per session scan A 75aa627cd676

Subscribe to this mod's changes

upstream-watch is a command published in the GitHub repository DougTrajano/pydantic-ai-skills (369 stars, last pushed 4d ago), licensed MIT. It adds 31 tokens to every session and 2,561 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.