g1t/crates/runner/src/learned.rs
| 1 | //! What the agent learned, asked for at the end of every run that changes |
| 2 | //! code and reported to g1t as memory candidates. |
| 3 | //! |
| 4 | //! The question rides on the prompt rather than costing another model call: |
| 5 | //! the agent ends its closing summary with a `<g1t-learned>` block of JSON, |
| 6 | //! which is taken out of the summary (so the pull request reads as before) |
| 7 | //! and sent with the run's own token. g1t keeps a candidate once a second, |
| 8 | //! independent source says the same, or a person keeps it. |
| 9 | |
| 10 | use std::time::Duration; |
| 11 | |
| 12 | use serde_json::{Value, json}; |
| 13 | |
| 14 | const OPEN: &str = "<g1t-learned>"; |
| 15 | const CLOSE: &str = "</g1t-learned>"; |
| 16 | /// The most items taken from one run. |
| 17 | const MAX_ITEMS: usize = 8; |
| 18 | const MAX_TEXT_CHARS: usize = 300; |
| 19 | const MAX_EVIDENCE_CHARS: usize = 300; |
| 20 | |
| 21 | /// What the agent is asked, after its task. |
| 22 | pub const ASK: &str = "When you are done, end your final message with what you learned here that the next agent would need and could not see at a glance: how to build or test, a convention, a decision and why, a trap. At most five short items, only ones you are sure of, never a secret, key or token. Write them as a JSON array inside <g1t-learned></g1t-learned> tags, each {\"kind\": \"fact\" | \"convention\" | \"decision\" | \"gotcha\", \"scope\": \"project\" | \"workspace\", \"text\": \"one sentence\", \"evidence\": \"what showed you, such as a command's output or a file\"}. scope workspace only for what holds in every project of the workspace. Write <g1t-learned>[]</g1t-learned> if there is nothing. The block is taken out of your summary."; |
| 23 | |
| 24 | /// The prompt with the question added. |
| 25 | pub fn ask(prompt: &str) -> String { |
| 26 | format!("{prompt}\n\n{ASK}") |
| 27 | } |
| 28 | |
| 29 | fn clip(text: &str, max: usize) -> String { |
| 30 | let text = text.split_whitespace().collect::<Vec<_>>().join(" "); |
| 31 | if text.chars().count() <= max { |
| 32 | text |
| 33 | } else { |
| 34 | text.chars().take(max).collect() |
| 35 | } |
| 36 | } |
| 37 | |
| 38 | /// The summary without its `<g1t-learned>` block, and the items in it. |
| 39 | /// A block that is missing or not JSON yields no items and leaves the |
| 40 | /// summary as it was, less the block. |
| 41 | pub fn split(summary: &str) -> (String, Vec<Value>) { |
| 42 | let Some(start) = summary.rfind(OPEN) else { |
| 43 | return (summary.trim().to_owned(), Vec::new()); |
| 44 | }; |
| 45 | let body_start = start + OPEN.len(); |
| 46 | let end = summary[body_start..].find(CLOSE).map(|at| body_start + at); |
| 47 | let body = &summary[body_start..end.unwrap_or(summary.len())]; |
| 48 | let rest = end.map_or("", |end| &summary[end + CLOSE.len()..]); |
| 49 | let clean = format!("{}{}", summary[..start].trim_end(), rest.trim_end()).trim().to_owned(); |
| 50 | let body = body.trim().trim_start_matches("```json").trim_start_matches("```").trim_end_matches("```").trim(); |
| 51 | let items = serde_json::from_str::<Vec<Value>>(body) |
| 52 | .unwrap_or_default() |
| 53 | .into_iter() |
| 54 | .filter_map(|item| { |
| 55 | let text = clip(item["text"].as_str()?, MAX_TEXT_CHARS); |
| 56 | if text.is_empty() { |
| 57 | return None; |
| 58 | } |
| 59 | let kind = item["kind"].as_str().filter(|kind| ["fact", "convention", "decision", "gotcha"].contains(kind)); |
| 60 | let scope = item["scope"].as_str().filter(|scope| ["project", "workspace"].contains(scope)); |
| 61 | Some(json!({ |
| 62 | "kind": kind.unwrap_or("fact"), |
| 63 | "scope": scope.unwrap_or("project"), |
| 64 | "text": text, |
| 65 | "evidence": item["evidence"].as_str().map(|evidence| clip(evidence, MAX_EVIDENCE_CHARS)), |
| 66 | })) |
| 67 | }) |
| 68 | .take(MAX_ITEMS) |
| 69 | .collect(); |
| 70 | (clean, items) |
| 71 | } |
| 72 | |
| 73 | /// Sends what was learned with the run's token. Never fails the run. |
| 74 | pub fn report(items: &[Value]) { |
| 75 | if items.is_empty() { |
| 76 | return; |
| 77 | } |
| 78 | let (Ok(run), Ok(token)) = (std::env::var("AGENT_RUN"), std::env::var("AGENT_RUN_TOKEN")) else { |
| 79 | return; |
| 80 | }; |
| 81 | let api = std::env::var("G1T_API").unwrap_or_else(|_| "https://api.g1t.sh".to_owned()); |
| 82 | let sent = ureq::post(&format!("{}/agent-runs/{run}/learned", api.trim_end_matches('/'))) |
| 83 | .timeout(Duration::from_secs(10)) |
| 84 | .send_json(json!({ "token": token, "items": items })); |
| 85 | if let Err(error) = sent { |
| 86 | eprintln!("g1t-runner: could not report what the agent learned: {error}"); |
| 87 | } |
| 88 | } |
| 89 | |
| 90 | /// The summary with its block taken out, after reporting what it held. |
| 91 | pub fn finish(summary: String) -> String { |
| 92 | let (clean, items) = split(&summary); |
| 93 | report(&items); |
| 94 | clean |
| 95 | } |
| 96 | |
| 97 | #[cfg(test)] |
| 98 | mod tests { |
| 99 | use super::*; |
| 100 | |
| 101 | #[test] |
| 102 | fn the_block_is_taken_out_and_read() { |
| 103 | let summary = "Fixed the date tests.\n\n<g1t-learned>\n[{\"kind\": \"gotcha\", \"scope\": \"project\", \"text\": \"The date tests need TZ=UTC.\", \"evidence\": \"npm test failed in CI time zone\"}, {\"kind\": \"nonsense\", \"text\": \"We use pnpm.\", \"scope\": \"workspace\"}, {\"kind\": \"fact\"}]\n</g1t-learned>"; |
| 104 | let (clean, items) = split(summary); |
| 105 | assert_eq!(clean, "Fixed the date tests."); |
| 106 | assert_eq!(items.len(), 2); |
| 107 | assert_eq!(items[0]["kind"], "gotcha"); |
| 108 | assert_eq!(items[0]["evidence"], "npm test failed in CI time zone"); |
| 109 | assert_eq!(items[1]["kind"], "fact"); |
| 110 | assert_eq!(items[1]["scope"], "workspace"); |
| 111 | } |
| 112 | |
| 113 | #[test] |
| 114 | fn a_missing_or_broken_block_teaches_nothing() { |
| 115 | assert_eq!(split("Done."), ("Done.".to_owned(), Vec::new())); |
| 116 | let (clean, items) = split("Done.\n<g1t-learned>not json</g1t-learned>"); |
| 117 | assert_eq!(clean, "Done."); |
| 118 | assert!(items.is_empty()); |
| 119 | let (clean, items) = split("Done.\n<g1t-learned>\n```json\n[]\n```"); |
| 120 | assert_eq!(clean, "Done."); |
| 121 | assert!(items.is_empty()); |
| 122 | } |
| 123 | |
| 124 | #[test] |
| 125 | fn the_question_follows_the_task() { |
| 126 | assert!(ask("Fix it.").starts_with("Fix it.\n\n")); |
| 127 | assert!(ASK.contains("never a secret")); |
| 128 | } |
| 129 | } |