magi-code 0.63.0

Repository-aware CLI coding agent for terminal work
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56

# Memory Benchmark History

Purpose: track peak RSS profiling runs over time for regression checks. Canonical workflow lives in [`docs/features/memory-usage-profiling.md`](memory-usage-profiling.md).

## Regenerate

```sh
python3 scripts/profile_memory.py --repo . --all --iterations 1
```

Artifacts written by default:

```text
target/profiling/memory/profile-results.json
target/profiling/memory/profile-summary.md
target/profiling/memory/<scenario>.log
```

Default run is offline and credential-free: temp `MC_HOME`, synthetic fixtures, no `auth.json`, and credential-looking env stripped from child processes.

## Run sources

| run | generated at | commit | source | notes |
| --- | --- | --- | --- | --- |
| Local baseline | unknown | unknown | `target/profiling/memory/profile-results.json` | Add first measured local run here; do not infer missing fields. |

## Scenario history

Primary memory metric is peak resident set size. Compare `max RSS MB` from same machine/profile only.

| scenario | run | elapsed ms | iterations | max RSS MB | primary metric | metric value |
| --- | --- | ---: | ---: | ---: | --- | ---: |
| startup_discovery | Local baseline | unknown | unknown | unknown | unknown | unknown |
| provider_sse_parser | Local baseline | unknown | unknown | unknown | unknown | unknown |
| rendering_heavy_transcript | Local baseline | unknown | unknown | unknown | unknown | unknown |
| tui_streaming_simulation | Local baseline | unknown | unknown | unknown | unknown | unknown |
| tool_timeout_cleanup | Local baseline | unknown | unknown | unknown | unknown | unknown |

## Delta rules

Use `max RSS MB` as main comparison value. For comparable runs:

```text
delta = (current max RSS MB - previous max RSS MB) / previous max RSS MB
```

Treat >5% peak RSS increase versus previous comparable run as investigation trigger, not automatic failure. If previous value is `unknown`, skip delta and collect a fresh baseline.

## Update rules

1. Run command above on same machine/profile when possible.
2. Copy `generated_at`, commit, platform, Rust version, and per-scenario rows from `profile-results.json`.
3. Add rows; do not overwrite history.
4. Record missing fields as `unknown`, never inferred.
5. Keep raw profiling artifacts out of git; default path is ignored under `target/`.
6. Record platform caveats when OS, allocator, Rust version, or release/debug profile differs from prior run.