Turn strava.py chat into a tool-using agent with runlog search

[sahilds1]

Description

Replace the passive single-shot chat with an agent loop: the conversation now runs llm's tool-calling loop (model.conversation(tools=[...]) + conversation.chain()), so the model can decide mid-conversation to call a tool and act on the result, instead of only answering from the seeded activity JSON.

The agent's first tool is search_runlog, a keyword search over the athlete's running log (RUNLOG.txt, override with --runlog). It lives in a new self-contained strava_tools.py built from pure parse_runlog / search_entries / format_entries helpers, which are the test targets in test_strava_tools.py. Adding a second tool is a one-line change to the tools=[...] list.

format_entries returns self-contained titles + bodies on purpose: with the default OpenAI (Chat Completions) plugin, reasoning state is not preserved across turns or across a tool hop, so the tool's returned text is the only durable state the next turn sees. strava.py carries a comment on this and on the OpenAI Agents SDK as a future extension path.

Also updates SYSTEM_PROMPT to tell the model about the tool and to ground history claims in what it returns, and documents the agent in CLAUDE.md.

Related Issue

Manual Tests

sahildshah•~/github/tiny-tools(2-turn-stravapy-into-an-agent⚡)» uv run strava.py --start 2026-07-11 --end 2026-07-15                            [18:56:15]
No rates present in response headers
Chatting about 4 activities. Ask a question ('exit' or Ctrl-D to quit).

> What are three things  that worked and one thing to improve



Based on the analysis of your recent activities, here are three things that worked and one area for improvement:

### Three Things That Worked:
1. **Consistent Run Performance:**
   - Your average speed across the last three runs was relatively consistent, ranging from 2.23 m/s (approximately 5:58 min/km) to 2.27 m/s (approximately 5:52 min/km) in your runs on July 11 and July 13. This consistency indicates a stable pacing strategy that you have managed to maintain.

2. **Effective Heart Rate Management:**
   - Your average heart rate varied (from 158 bpm to 178 bpm) but is reflective of a well-controlled effort. Keeping your heart rate around 150-180 bpm during runs suggests you're working at a good intensity without overexerting yourself, particularly during longer distances.

3. **Best Efforts Indication:**
   - You've shown impressive best efforts in shorter distances (400m, 1K, mile segments) with good times recorded, indicating that your speed and anaerobic capacity are on the rise. For example, your 400m best effort was completed in 159 seconds during your July 11 run, suggesting strong speed work during your training.

### One Thing to Improve:
- **Elevation Gain and Adaptation:**
  - Your runs have come with minimal elevation gain (e.g., 57.6m on July 11 and 76.6m on July 13), suggesting you might benefit from incorporating hill workouts. Adding elevation can improve strength and power during runs. Focus on gradual increases in hill sessions once a week to enhance your overall performance and adapt your leg muscles to varying terrain.

Incorporating these insights into your training plan can help refine your approach and maximize your performance in future workouts. Let me know if you have any further questions or would like more specific strategies!

>
>
>
> Why didn't I run on  Wednesday July 15 2026
According to the Strava activities data, you did not have any recorded running activity on Wednesday, July 15, 2026. The only activity logged on that date was a weight training session, which lasted 30 minutes (1823 seconds), and there was no mention or record of a running activity.

It's possible that you may have scheduled a rest day, opted for cross-training (like the weight training), or simply missed your run for any number of personal reasons (time constraints, fatigue, or recovery needs).

If you're looking for a more specific reason or pattern in your training, let me know, and I can check your past logs for more context!

> Check my past logsf or more context
The search of your past logs indicates that on **Wednesday, July 15, 2026**, there was a **heat advisory** in effect from 10 AM to PM ET. This weather condition may have influenced your decision to skip running that day, prioritizing safety and avoiding heat-related issues.

Additionally, your last recorded run prior to that was on **Tuesday, July 14, 2026**, where you mentioned kicking at the end of the run, indicating you had a solid training session just before the heat advisory.

In summary, it appears that the combination of the heat advisory and possibly needing a rest day after a solid run contributed to your decision not to run on July 15.

Automated Tests

Documentation

Reviewers

Notes